Alkali-Silica Reactions in Concrete: David Stark

34
Alkali-Silica Reactions in Concrete
David Stark1
Preface decades, became identified in most areas of the world. The fact
that aggregates must be evaluated for chemical stability in con-
THIS CHAPTER IS THE FOURTH OF A SERIES DEALING crete as well as for physical competence thus became evident
with chemical reactions of aggregates in concrete. The first ver- with costly realization. In response to this situation, ASTM de-
sion, entitled Chemical Reactions, was written by W. C. veloped standards to identify and evaluate construction mate-
Hansen and was included in ASTM STP 169A, published in rials to avoid deleterious ASR. Salient features of ASR, together
1966. It dealt with several types of chemical reactions known at with a summary of methods of preventing ASR, are given in this
the time to occur in concrete, including alkali-silica reactivity. chapter.
The second version was written in 1978 by Sidney Diamond for
ASTM STP 169B and was entitled Chemical Reactions Other Symptoms of Alkali-Silica Reactivity
Than Carbonate Reactions. It specifically excluded alkali-car-
bonate reactivity and emphasized alkali-silica reactivity. The ASR may develop in hardened concrete and may appear with
third version, written by the present author, covers only Alkali- one or more characteristics [2]. Abnormal cracking may develop
Silica Reactions in Concrete. This fourth edition reviews and in a variety of patterns, depending on the design of the structure,
updates the topics as addressed in cement and concrete articles reinforcement, detailing, restraints, and exposure conditions, as
and introduces new technology that has been developed. shown in Figs. 13. Figure 1 illustrates typical fine pattern crack-
The shifting emphasis in this chapter since 1966 reflects ing due to ASR. Here, cracks are generally random in many
the predominating concern with alkali-silica reactivity in pres- directions and with little, if any, preferred directional restraint
ent-day construction and reflects our updated knowledge of to expansion. Figure 2 illustrates well-defined longitudinal crack-
how to deal more effectively with the problem. Accordingly, ing along a pavement highway, caused by ASR. Here, maximum
more detail and emphasis than previously included are pro- restraint to expansion extends parallel to the longitudinal pave-
vided for the practicing technologist responsible for providing ment direction, thereby resulting in greater expansion in the
long-term durable concrete. transverse direction of the pavement. Figure 3 reveals cracking
in a generally horizontal plane in which little restraint develops
Introduction in the vertical direction. The important point is that the greatest
cracking due to ASR can develop in the direction of least resist-
The selection of aggregates for use in concrete was based pri- ance, provided uniform restraint on the concrete does not
marily on physical characteristics such as grading, particle exceed the tensile strength in the concrete structure.
shape, hardness, density, and cleanness. Such characteristics Characteristic features resulting from ASR are shown in
were the major, if not the only, consideration of the technolo- Fig. 4. They include reaction rims in peripheries of aggregate
gist in achieving long-term specified concrete strengths and particles, microcracks in the concrete, and ASR gel in the mi-
volume stability. Virtually no attention was given to the chem- crocracks. It should be noted that rims may be due only to
ical or mineralogical composition of the aggregate, despite the weathering in natural gravels and, as such, are not necessarily
known fact that concrete is a highly alkaline system in which related to ASR.
pore solutions usually exceed pH values of 13. The best evidence on which to confirm the development of
In the late 1930s, Stanton, of the California Division of ASR is the presence of alkali silica gel. It is a reaction product
Highways, detected a previously unknown deleterious chemi- of alkaline solutions in the concrete and reactive forms of silica
cal reaction involving pore solutions in the concrete and certain or siliceous components of the aggregate, which swells and
compositions of siliceous aggregates [1]. He determined that expands with the uptake of moisture. It should be emphasized
resulting expansion could lead to abnormal cracking, reduction that simply the presence of gel deposits does not confirm the
in strength, and early in-service failure of the concrete structure. occurrence of associated distress. Rather, it denotes only that the
This phenomenon became known as alkali-silica reactivity reaction had developed.
(ASR). The presence of alkali silica gel can be confirmed by de-
By the 1940s, ASR was known to have developed in wide- termination of refractive indices using the petrographic mi-
spread areas of the United States, and through the following croscope, together with applying uranium acetate solution and
1
CTL Group, 5400 Old Orchard Road, Skokie, IL.
401
STARK ON ALKALI-SILICA REACTIONS 402
Fig. 1Fine pattern or random cracking initiated by ASR in a parapet wall of a

dam. There is little, if any, cracking relating to differential restraint to expansion. It
may be somewhat accentuated by drying shrinkage.
viewing using ultraviolet light in the stereomicroscope, as de- Each of these factors variably can affect concrete per-
scribed in the annex of C 856 [2]. formance as noted below:
1. Hydroxyl ion concentration in the concrete solution is the
Mechanism of Reactions and Distress in crucial factor in potential for ASR, not the amount of
Concrete sodium or potassium in, for example, cementitious mate-
rials.
Voluminous literature [3,4] has been produced dealing with all 2. Numerous siliceous aggregate materials are not equally
aspects of ASR. In its simplest and most straightforward form, deleteriously reactive. For example, opal and volcanic tuff
the reaction can be described as a two-step process as follows: materials are often highly reactive while certain chert or
Step 1Aggregate particle + alkali hydroxide solution silicate minerals are essentially innocuous.
Gel reaction product 3. ASR gel may exert variable pressure and expansion in the
Step 2Gel reaction product + Water Swelling and pos- concrete, depending on composition, viscosity, and mois-
sible expansion ture availability.
Fig. 2Longitudinal cracking initiated by ASR in a jointed, slab-on-grade

pavement. A longitudinal crack pattern prevails in this pavement, together with
minor random cracking, particularly near transverse joints.
Overall, the Powers and Steinour [5] model best fits pub-
lished results. They proposed that safe reactions are those in
which sufficient calcium ion is available in alkaline pore solu-
tions to produce limited-swelling high-calcium-content gel. Con-
versely, when less calcium ion is available, a higher alkali con-
tent gel forms, which has a greater capacity to swell with
absorption of moisture.
The initial surface reaction in this model is assumed to oc-
cur in contact with the pore solution in the cement paste, which
is saturated with respect to calcium hydroxide. It therefore pro-
duces essentially a non-swelling reaction product. For a safe re-
action to continue, it is assumed that consumption of alkali in
the reaction results in greater solubility and availability of
calcium at subsequent reaction sites, thereby continuing to pro-
duce a high calcium non-swelling gel reaction product. These
safe reactions were believed to begin and continue without ex-
pansion if the initial alkali concentration is not greater than that
produced by cements of about 0.6 % equivalent sodium oxide
(Na2O), or about 0.4 normal sodium hydroxide (0.4N NaOH)
concentration in pore solutions at 0.50 water-cement ratio (w/c)
by mass.
According to Powers and Steinour, an expansive reaction
occurs when alkali concentrations in pore solutions are higher,
thus reducing the concentration of calcium ion. Accordingly, a
low-calcium, high-alkali unlimited-swelling gel reaction product
is formed. In this case, the reacted layer in the aggregate be-
comes too thick, and the calcium ion cannot diffuse to the re-
action site at a sufficient rate to prevent formation of a swelling
gel. The diameter of the hydrated calcium ion is substantially
larger than that of the hydrated sodium or potassium ion, with
the result that they diffuse more readily than the calcium ion to
the internal reaction sites within a particle of reactive ag-
Fig. 3Vertical concrete core from jointed pavement in gregate. The lower solubility of calcium in solution and the
which major cracking due to ASR developed primarily in a lower diffusion rate of calcium to reaction sites within the par-
horizontal orientation. Lesser cracking is evident vertically and
ticle both serve to reduce its participation in the reaction. How-
diagonally. The horizontal cracking developed normal to the
general direction of least resistance to cracking.
ever, if the amount of reactive silica, or its fineness, is increased
Fig. 4Evidence of ASR in sample of concrete from gravity dam. Coarse

aggregate particles were quarried and composed of shistose to mylonitic
composition with the largest aggregate particle having a maximum size of 2 in.
ASR-related features include well-defined, dark reaction rims and microcracks
through aggregate particles and cement paste matrix that contain white-
colored ASR gel deposits.
sufficiently, the alkali content of the solution is rapidly reduced pumice, which are very dense or very porous volcanic glasses, re-
to a safe level so that a higher calcium, limited-swelling gel can spectively, also are known to have reacted deleteriously in con-
form. Obviously, intermediate-alkali and corresponding calcium crete. Slightly metamorphosed volcanic rocks, such as metarhy-
ion concentrations, diffusion rates, permeabilities of reactive olite, are also known to produce expansive ASR.
aggregates, and moisture availabilities exist and result in
intermediate conditions in which ASR develops. Moisture Availability and Environmental Effects
Another aspect of the Powers and Steinour model is the As indicated, moisture must be sufficiently available in the con-
diffusion of silica away from reaction sites and out of reacted crete to permit expansive ASR to occur. Laboratory data indi-
particles. They proposed that silica was able to diffuse out cate that relative humidity (RH) values in the concrete must ex-
through the reaction layer in the aggregate particle and that ceed about 80 %, referenced to 23 1C, before expansion due
expansion-controlling processes were diffusion rates of alkali to ASR can develop [9]. This condition is easily met in outdoor
and calcium to reaction sites. Chatterji [6] suggested that slab-on-grade concrete, regardless of climate, including desert
calcium hydroxide plays an additional role in ASR in that it can conditions. Interior portions of elevated concrete members,
cause or increase expansions by impeding the escape of dis- such as bridge decks, support columns, and girders, also ex-
solved silica from reactive particles. In this circumstance, cal- ceed the 80 % RH level, at least on a cyclic basis, even where
cium reacts to form a calcium-alkali silica gel that blocks the prolonged severe drying conditions prevail. Except for near-
escape of other reaction products out of the particle. surface concrete in elevated structures, concrete climates can
ASR gels vary widely in composition, depending not only be expected to be sufficiently damp to support expansive ASR
on circumstances just indicated, but on other factors such as on an almost continuous basis. Near-surface concrete experi-
distance from the reaction site and movement along cracks [7]. ences major RH fluctuations that support expansive ASR on an
Electron microprobe and X-ray studies suggest that ASR gels intermittent basis, regardless of climate. Where high water-
are two-phase composites consisting of the swelling alkali- cement ratios are used, residual mixing water may be sufficient
silicate hydrate phase of relatively narrow composition and a to support expansive ASR in mass concrete, even in interior air-
limited swelling calcium-alkali silicate hydrate phase, which conditioned exposures.
varies in both proportion in the gel and composition. Being cal-
cium-bearing, it provides some rigidity to the paste structure. Methods of Identifying Potentially
Viscosity of the gel is a major factor affecting swelling pres- Reactive Aggregate
sures generated in concrete.
ASTM has five standards available to identify potentially reac-
Types of Reactive Aggregate tive siliceous aggregates. They are:
Potentially deleteriously reactive rock types probably exist in 1. ASTM Standard Guide for Petrographic Examination of
every state in the United States and in most countries through- Aggregates for Concrete (C 295)
out the world. Many natural siliceous rock types are known to 2. ASTM Standard Test Method for Potential Alkali-Silica
have reacted deleteriously with the highly alkaline solutions in Reactivity of Aggregates (Chemical Method) (C 289)
concrete. Probably the most widely used of such aggregates are 3. ASTM Standard Test Method for Potential Alkali Reactivity
those that contain noncrystalline or poorly crystalline silica. of Cement-Aggregate Combinations (Mortar-Bar Method) (C
These include opaline and chalcedonic silica that are found in 227)
such rock types as chert and flint, as secondary fillings in voids 4. ASTM Standard Test Method for Potential Alkali Reactivity
in, for example, basalt, and as interstitial and interlayer material of Aggregates (Mortar Bar Method) (C 1260)
in rock types such as sandstone. Reactive silica of this type also 5. ASTM Standard Test Method for Determination of Length
is found as microscopic veinlets or finely disseminated in other- Change of Concrete Due to Alkali-Silica Reaction (C 1293)
wise innocuous rock types such as limestone and dolomite. The Another procedure, ASTM Standard Test Method for
most highly reactive rock types known to the writer are shales Potential Volume Change of Cement-Aggregate Combinations
that contain opaline diatoms, which have reacted deleteriously (C 342), was intended to identify expansive cement-aggregate
and produce popouts within one day of placing the concrete [8]. reactions involving so-called sand-gravel aggregates that occur
Deleteriously reactive silica also is present in metamorphic primarily in eastern Colorado, Kansas, Nebraska, Iowa, Mis-
rocks where it is classified as strained quartz and microcrys- souri, and southeastern Wyoming. The basis for this test was
talline quartz. In these cases, crystal structures are variously dis- the belief that the cause of distress observed in concrete made
torted, thus rendering them susceptible to deleterious ASR. This with these aggregates was more or less unique to those areas
imperfectly crystallized quartz has been recognized as the re- and so was given a separate test procedure. The author believes
active component of such metamorphic rocks as gneiss, schist, that alkali-silica reactivity is the major cause of distress in those
metagraywacke, and quartzite. These rock types are slowly re- areas and that the test is of little additional value. This standard
active and are known to have exhibited deleterious reactivity was discontinued in 2001.
with high-alkali cements only after five or more years of service.
The other general reactive component of siliceous aggregates ASTM C 295Petrographic Examination
is the glassy to cryptocrystalline matrix of volcanic rocks of of Aggregates
approximately rhyolitic to andesitic composition. This mate- ASTM C 295, together with ASTM C 294, Standard Descriptive
rial, especially tuff, constitutes a major proportion of these rock Nomenclature for Constituents of Concrete Aggregates, is a
types and is commonly altered by weathering that facilitates pen- guide recommended for identifying and describing con-
etration of alkaline solutions into the aggregate particle in the stituents of aggregate by petrographic examination of concrete.
concrete. Weathered volcanic rock is particularly troublesome It can be done rapidly but is somewhat subjective in nature, de-
in that it has produced ASR-associated distress when used with pending on the experience and capabilities of the petrographer
cements with less than 0.60 % equivalent Na 2O. Obsidian and examining the aggregate. Basically, results depend on the
judgment of the petrographer in contrast to fixed measuring for the duration of the test period. A zero reference comparator
procedures for classifying aggregate. A major use is for alerting reading was made the day following casting, then biweekly to
materials engineers to the presence of potentially reactive rock bimonthly to a suggested test period of six months.
and mineral types, and is highly recommended for use in The Appendix of the ASTM Standard Specification for Con-
conjunction with other tests and field performance evaluations. crete Aggregates (C 33) states that in C 227, cement-aggregate
combinations showing expansions greater than 0.10 % at six
ASTM C 289Chemical Method months usually should be considered capable of harmful reac-
One procedure used to identify whether an aggregate is poten- tivity. The Appendix also states that combinations showing ex-
tially deleteriously reactive is ASTM C 289, commonly referred pansions of at least 0.05 % at three months should be considered
to as the Quick Chemical Test. This method was originally potentially capable of harmful reactivity. During the 1950s, the
developed by Mielenz and co-workers at the U.S. Bureau of 0.05 % and 0.10 % expansion levels were applied to test ages of
Reclamation [10]. In this procedure, the aggregate in question is six and twelve months, respectively. This then was a stricter test
sized to the 300150 micron sieves, then three 25-g samples of criterion. Today, the 0.05 % criterion at three months is consid-
the material are immersed in 1N NaOH solution in sealed ered innocuous if the six months criterion is less than 0.10 %.
containers held at 80 1C for 24 h. The resulting suspensions Many years of experience with C 227 have revealed short-
are then filtered, and the filtrates are analyzed for reduction in comings that, in some cases, have resulted in misleading con-
the original hydroxyl ion concentration caused by reaction with clusions with eventual deterioration of concrete. The suggested
the test sample and for the amount of silica dissolved in the expansion criteria appear to be valid for rock or mineral types
solution. The result is plotted on a graph where dissolved silica arbitrarily defined as rapidly reactive aggregates, such as opa-
(Sc) in millimoles per liter is plotted logarithmically on the line-bearing materials and weathered glassy to cryptocrys-
abscissa and reduction in alkalinity (Rc) is plotted on a linear talline volcanics of rhyolitic to andesitic composition, when they
scale on the ordinate. A curve is drawn on the graph in which are tested with cement with 1.0 % alkali. C 33 leaves open to the
Aggregate Considered Innocuous is located on one side of the engineers discretion the option of extending the test period to,
curve, and Aggregate Considered Deleterious or Potentially for example, 12 or more months. However, extending the test
Deleterious is located on the other. Innocuous aggregates are period to more than 12 months still may result in failure to
considered to be those that produce little or no reduction in define so-called slowly reactive aggregates as potentially
alkalinity, or those for which a significant reduction in deleteriously reactive in field structures. These aggregate types
alkalinity is accompanied by relatively little dissolution of silica. include certain gneisses, schists, graywackes, and metavol-
Location of the innocuous-deleterious curve on the graph canics. Even when used with cements with 1.0 % alkali, expan-
presumably is based on field performance and on ASTM C 227 sions for these rock types may reach no more than 0.05 % after
mortar-bar test results, with the mortar bar criterion being 0.10 two to three, or more, years [11]. Such a timeframe is not
% or greater linear expansion. practical in these cases.
ASTM C 289 appears to have some serious limitations and A further limitation from C 227 results is a recommenda-
is seldom recommended by various organizations and coun- tion to use low-alkali cement (less than 0.60 % equivalent Na2O)
tries in their practices and standards. It fails to identify slowly if an aggregate is judged to be potentially deleteriously reac-
reactive aggregates, such as gneiss, schist, or quartzite, where tive. Field and laboratory studies [11] have revealed that
reactivity results from strained or microcrystalline quartz. certain aggregates, such as glassy volcanics or tuffs, are delete-
Furthermore, reports have shown that mineral phases, such as riously reactive with low as well as high-alkali cements.
calcium, magnesium, iron carbonates, gypsum, and zeolites, Leaching of alkalies from the concrete also has been found
interfere in the results of the test, thereby underestimating Sc to be a serious limitation on expansion [12,13] in C 227. An
values or overestimating Rc values, and in the classification of aggregate may be judged to be innocuous when used with
known reactive aggregates as innocuous. On the other hand, the cement of a certain alkali level when, in fact, it may be delete-
test was found to be severe for several aggregates with good riously reactive.
field service records where it was attributed to the fine particle Overall, C 227 may be useful for identifying certain aggre-
size of the material (150300 m) and high test temperature gates as deleteriously reactive. However, many cases exist
(80C). where the procedure falls short of expectations.
Overall, the test could be used as a quality control tool for
aggregate procedures only if results of the test are confirmed ASTM C 1260Potential Alkali Reactivity of
by more reliable laboratory tests or field performance records. Aggregates
It is evident that limitations exist in ASTM C 227 and C 289 for
ASTM C 227Potential Alkali Reactivity identifying known slowly reactive aggregates such as gneiss,
of Cement-Aggregate Combinations graywacke, some siliceous limestones, schist, and quartzite.
ASTM C 227 has been an ASTM mortar bar standard since 1950 Thus, a test method was developed in South Africa at the Na-
and was initially conceived by Stanton as a means of identify- tional Building Research Institute [14] and was designed to
ing safe or unsafe cement-aggregate combinations. As such, it rapidly identify slowly and highly reactive aggregates, based on
simulated what might actually occur in concrete [1]. Testing was field performance record. With some modifications in the late
carried out using the aggregate in question, which was sized to 1980s, it became an ASTM standard in 1990 and is known as
a fine aggregate grading. If a coarse aggregate was to be the Potential Alkali Reactivity of Aggregates (Mortar Bar
evaluated, it had to be crushed, sized, and washed to the pre- Method).
scribed grading as well. A high-alkali cement, greater than 0.60 For this method, the aggregate is sized and washed to meet
% as equivalent Na2O, or the contemplated job cement, was to the specified grading, then used at a proportion of one part of
be used in the test. Four companion mortar bars were made, cement and 2.25 parts of aggregate, then cast into mortar bars of
then stored over water in a sealed container held at 38 2C 0.47 water-cement ratio by mass, and stored in a moist
cabinet at 23 2C. One day after casting, the specimens are NaOH solution is much greater than most alkali levels in field
removed from the molds, a comparator reading is obtained, concrete and therefore cannot provide realistic safe cement
then it is stored for 24 h in water brought to 80 2C. An ad- alkali levels for the aggregate tested.
ditional comparator reading is obtained, then the bars are To help resolve any of these issues, recent work has shown
transferred to 1N NaOH solution and again stored at 80 2C the potential for an accelerated version of the test to be per-
in sealed containers for 14 days. Periodic comparator readings, formed at 60C at greater than 95 % relative humidity over a
for example, at 1, 3, 7, 10, and 14 days, are made, while being three-month period. Earlier, Stark et al. [15] used fine and
careful to prevent cooling and drying while taking the read- coarse aggregate, received as concrete aggregate. Tests of con-
ings. Total time required between making the mortar bars and crete prisms up to 42 days in 1N NaOH solution at 80C showed
obtaining the final measurement of length change is 16 days. some success. Other specimens were cast with the same alkali
Test criteria suggested in the Appendix of C 33 are as follows: level as that used for the immersion test, also with some success.
Expansions in NaOH solution greater than 0.20 % at 16 Overall, none of the existing ASTM test methods provide
days indicate potentially deleterious expansion. cement alkali levels and environments that can reproduce
Expansions between 0.10 % and 0.20 % are uncertain but given cement-aggregate combinations that exactly prevent or
are known to include aggregates that can be deleterious or allow deleterious ASR. Existing tests are either too lenient, too
innocuous in field concrete. severe, or too time-consuming. One rapid procedure that has
Expansions not greater than 0.10 % represent innocuous been developed, and can include mineral admixtures, is not yet
aggregate. an ASTM Standard [16].
Several important points must be emphasized in this pro-
cedure. First, the alkali content of the cement used in making the Avoiding Expansive Alkali-Silica Reaction
mortar bars may be comparatively low because a much greater
NaOH concentration is used in the immersion solution and Three methods are used today to prevent abnormal expansions
diffuses into the mortar bars, thus exerting overwhelming due to ASR. They are:
control on reactions. Second, the test result indicates only that 1. Avoid the use of reactive aggregate.
the aggregate may be potentially reactive, not that it will react 2. Limit the cement alkali level.
deleteriously in field concrete. Cement alkali level and 3. Use pozzolans or other admixtures in the fresh concrete.
environment exert control on field performance. Third, the test The most feasible method for a particular job will depend
has successfully identified, in confor- mance with field on economics and the local availability of suitable materials. In
observations, the deleterious nature of reactivity with most some cases, combinations of the various methods have been
slowly reactive aggregates, such as certain gneisses, schists, used. Each of the three methods is discussed below.
and quartzites. Overall, this procedure is rapid but may
overstate the shortcomings found in ASTM C 227 and C 289. Use Innocuous Aggregate
At first glance, it would seem a simple matter to use innocuous
ASTM C 1293Length Change of Concrete Due aggregates to avoid expansive ASR. However, this presumes
to Alkali-Silica Reaction that the innocuous aggregate has a proven service record and
ASTM C 1293 is considered by some to be the most reliable has been evaluated properly using pertinent ASTM standards.
procedure among ASTM test methods to evaluate aggregates If a field service record is to be used, assurance must be made
for alkali silica reaction. It allows for the evaluation of aggre- that the prospective aggregate from a commercial or lithologic
gates as they are used in concrete. Coarse aggregate can be source is of similar petrographic character as that on which
evaluated without crushing them, or fine aggregates can be the service record is based. If quarried stone is being consid-
tested without sieving and recombining according to pre- ered, for example, one must be careful to evaluate the same
scribed particle size distributions. It also is the only method rock strata, particularly where bedding is convoluted or
where combinations of coarse and fine aggregates proposed inclined. If water-laid deposits from a given source of sand or
for use in job applications can be tested. gravel are being considered, one must be sure to evaluate
In general, the test is effective in recognizing the potential materials from, for example, the same river terrace level.
alkali silica reactivity of both the rapidly reactive aggregates, Petrologic character of the source material can vary signifi-
such as opaline-bearing, glassy cryptocrystalline volcanic rocks, cantly not only vertically from bed to bed, but also laterally.
as well as the slowly reactive aggregates such as granite gneiss, Variations often develop in mineralogic composition, propor-
graywackes, schists, and some sandstones and limestones. tion, and particle size of individual rock types in the prospec-
In this procedure, equivalent alkali in the cement plus tive aggregate source. Use of an alternate source of a durable
NaOH solution added to the mixture should be 1.25 % as Na2O, aggregate may result in major transportation costs for
and the w/c should range between 0.42 and 0.45. Three com- shipment to the job site.
panion concrete prisms are cast, cured, and stored in a moist The use of field service record also requires that cement
environment in a sealed container at 38 2C for up to 12 alkali level be known for the existing concrete, regardless of
months or more as needed. C 33 indicates the concrete would whether low or high alkali cement was used previously and will
be considered deleterious if expansion reaches 0.040 % or be used in new construction. An aggregate that is known to be
more at 12 months. slowly reactive, when used with a more rapidly reactive aggre-
Several limitations exist in the C 1293 standard. First, it gate, may then become innocuous due to the remaining lower
may require a year or more to obtain a meaningful result, concentration of alkali in solution in the concrete.
depending on the nature of the aggregate in question. Second, an A second major factor in evaluating field service record is
alkali level of 1.25 % as equivalent Na2O of the cement plus the environment to which the concrete is exposed. This in-
cludes temperature, moisture accessibility, and restraint.
Elevated temperatures and ready availability of moisture are
known to exacerbate alkali-silica reactivity, whereas restraint Use Admixed Materials

or confining pressure reduces expansion and reactivity- Certain finely divided ground materials may be added to con-
induced distress. crete to permit the safe use of otherwise potentially reactive ce-
Deleterious ASR has been observed in concrete structures ment-aggregate combinations. These materials are intended to
in all habitable climates, including hot and cool damp regions react in the normal course of cement hydration by complexing
where the concrete is exposed directly to the ambient outdoor alkalies to reduce hydroxyl ion concentrations in concrete, or by
conditions. Moisture usually is sufficiently available, either altering diffusion rates of alkali and lime to reaction sites. This
continuously or cyclically, to support expansive ASR, and that results in the formation of harmless nonexpansive reaction
temperature change, in a practical sense, serves only to change products. The most widely used material for this purpose is fly
the rate of reaction. Commonly, increases in ambient outdoor ash obtained from power plants burning coal. Other, less
temperature are accompanied by reductions in relative commonly used materials include ground granulated blast-
humidity with possible offsetting effects on ASR. For example, furnace slag, finely ground volcanic glass, and silica fume. In
increase in temperature may increase the rate of chemical re- addition, lithium solutions, which have been found to be ex-
action, but drying would remove water that might otherwise be tremely effective in reducing expansions in experimental work,
available for absorption by and resulting expansion of ASR gel. are contemplated for the future. The use and testing of these
Water-cement ratio also will affect reactivity, with residual mix- materials are discussed below.
ing water possibly being the only source of moisture available
to support expansive reactions. Fly Ash
A third factor that has been attributed to field performance Two categories of fly ash are recognized in ASTM Standard
is extraneous sources of alkali, specifically, hydroxyl ion. Specification for Coal Fly Ash and Raw or Calcined Natural
Examples include sodium chloride (NaCl) deicer salts, alkali Pozzolan for Use in Concrete (C 618): Class F and Class C. For
salts occurring naturally in soils [e.g., sodium sulfate (Na 2SO4], purposes of preventing expansive ASR, the most important dis-
industrial solutions (e.g., NaOH), and seawater. Calculations tinction is the minimum permissible proportion of SiO2 + A12O3
suggest that many times the quantity of alkali present in the ce- + Fe2O3, which for Class F ash is 70 % and for Class C ash is
ment can diffuse into concrete from external sources, possibly 50 %. Class C ashes usually have greater lime (CaO) contents,
increase hydroxyl ion concentration, and perhaps convert an often exceeding 10 % and even 20 %. High lime contents
otherwise innocuous cement-aggregate combination into a adversely affect the capability of a fly ash to prevent expansive
deleterious one. Inspection must consider all such factors in any reactivity [19]. An optional requirement of fly ash is a 1.5 %
assessment of field performance relating to potential for maximum limit on available alkalies as determined by meth-
deleterious ASR. ods described in Standard Test Methods for Sampling and Test-
Results of laboratory research have alluded to so-called ing Fly Ash or Natural Pozzolans for Use in Portland Cement
pessimum (intermediate) proportions of reactive aggregate at Concrete (C 311). The latter restriction is important because
which maximum expansion rates and levels develop [5]. This total alkali content of cement + fly ash may result in the equiv-
appears to be largely of academic interest except to explain alent of very high alkali levels that may convert an otherwise
some case histories in which all of the aggregate was reactive innocuous cement-aggregate combination into a deleterious one
but no damage resulted, while in related cases only a relatively [20].
small proportion of reactive material was used and damage re- ASTM provides several testing procedures to determine the
sulted. Attempting to specify aggregates for field concrete on suitability of a fly ash for controlling expansive ASR in con-
the basis of avoiding pessimum ratios should be avoided. crete. The Standard Test Method for Effectiveness of Pozzolans
or Ground Blast-Furnace Slag in Preventing Excessive Expan-
Limit Cement Alkali Level sion of Concrete Due to the Alkali-Silica Reaction (C 441) in-
As noted earlier, cement alkali level is a major factor deter- volves replacing 25 % of the cement, by mass, with an amount
mining expansion due to ASR. In general, the higher the ce- of fly ash equal to the volume of cement replaced. The alkali
ment alkali level, the greater will be expansion due to reactiv- content of the cement should be about 1.0 %. Highly reactive
ity. Simply specifying low-alkali cement (less than 0.60 % Pyrex glass is used as fine aggregate in a specified grading, and
equivalent Na2O) to avoid expansive ASR has not always been the water content is gauged to meet a flow requirement. Com-
successful. In some cases, cement alkali levels as low as 0.45 panion mortars are made with and without the fly ash. Test
0.50 % or less have failed to prevent expansive reactivity where storage is over water in sealed containers held at 38 2C. Ex-
partially weathered glassy to cryptocrystalline volcanic rock of pansion using the fly ash is determined at 14 days, and the ab-
rhyolitic to andesitic composition constitute no more than 5 % solute expansion level of the test mortar is evaluated.
of the aggregate [11]. In this regard, alkalies from, for example, ASTM C 618 requires expansion of the test mixture to not
feldspars or noncrystalline volcanic aggregates can be released exceed 100 % at 14 days when using low alkali cement.
by water or calcium hydroxide solutions, thus resulting in the C 441 has been criticized for a number of reasons. First,
development of expansive ASR. The best indicator presently specific Pyrex glass is used as the standard reactive aggregate.
available to determine a safe cement alkali level is field It contains appreciable alkali that, it has been argued, inter- feres
performance record where all pertinent factors are properly with proper evaluation of the pozzolan. Of much greater
considered. importance is the fact that Pyrex glass reacts significantly
Deleterious ASR also can occur in concrete in which ca- faster than many fly ashes, while in field concrete containing
thodic connection or stray current systems develop [17,18]. It natural aggregate, the reverse is probably true. For this same
appeared that current densities of at least 25mA/m2 caused the reason, meeting the C 618 expansion criterion at 14 days for the
greatest expansion when the cement alkali level was high. How- test mortar will be difficult, if not impossible, to meet, even
ever, there was no expansion when the cement alkali level was though the fly ash might effectively suppress expansive alkali-
low. silica reaction.
Raw or Calcined Natural Pozzolan An important advantage of lithium salts over other ad-
Remarks presented for fly ash apply generally to raw or cal- mixed materials is the fixed composition of the salt, thus re-
cined natural pozzolans, such as some volcanic ashes, opaline quiring no testing for individual concrete mix materials [21].
shales, and diatomaceous earths. However, C 618 recommends ASTM presently does not address this means of preventing
75 % reduction in expansion after 14 days under C 441 test expansive ASR. A very beneficial but, today, not practical use for
conditions, as noted earlier. lithium solution would be its penetration to uncracked hard-
ened concrete to prevent or mitigate distress due to ASR [27].
Ground Granulated Blast-Furnace Slag
(GGBFS) Concluding Remarks
Ground granulated blast-furnace slag can be used to prevent
expansive ASR. C 441 is used to evaluate slag as well as fly ash, The concrete technologist has at his disposal broad knowl-
except that 50 % by mass of portland cement is replaced by slag edge and guidelines to produce concrete that is not susceptible
equal to the volume of cement replaced. ASTM Standard to deleterious ASR. ASTM provides various test procedures and
Specification for Ground Granulated Blast-Furnace Slag for Use recommendations that, if used prudently with other informa-
in Concrete and Mortar (C 989) suggests that reduction in tion noted in this chapter, should ensure that deleterious reac-
expansion should be at least 75 %. tions can be avoided in future construction.
The mechanism by which slags inhibit expansive reactivity

is purported to be different than that by which fly ashes prevent References
expansion. From laboratory investigations [21], it was
[1] Stanton, T., Expansion of Concrete Through Reaction between
concluded that slag hydration in portland cement-water systems
Cement and Aggregate, American Society of Civil Engineers,
serves to reduce diffusion rates of alkali to aggregate reaction New York, Dec. 1940, pp. 17811811.
sites, thereby minimizing the rate of ASR. This would have the [2] Stark, D., Handbook for The Identification of Alkali-Silica Reac-
effect of producing a potentially less expansive or nonexpansive tivity In Highway Structures, SHRP-C/FR-91-101, Strategic High-
gel reaction product. This explanation needs further way Research Program, National Research Council, Washing- ton,
confirmation. DC, 1991.
[3] Diamond, S., Alkali Aggregate Reactions in Concrete: An
Silica Fume Annotated Bibliography 19391991, Strategic Highway
Silica fume is occasionally used to prevent expansive ASR. It Research Program, SHRP C/VWP-92-601, National Research
appears to function in much the same manner as fly ash in re- Council, Washington, DC, 1992.
ducing expansions due to reactivity. That is, the calcium silicate [4] Helmuth, R., Alkali-Silica Reactivity in ConcreteAn Overview
hydrate reaction product complexes alkali, thereby reducing the of Research, Strategic Highway Research Program, SHRP-C342,
National Research Council, Washington, DC, 1993, 105 pp.
hydroxyl-ion concentration in the pore solution in the concrete.
[5] Powers, T. C. and Steinour, H., An Interpretation of Published
Standard Specification for Use of Silica Fume as a Min- eral
Researches on the Alkali-Aggregate Reaction, Journal,
Admixture in Hydraulic Cement Concrete, Mortar, and Grout (C American Concrete Institute; Proceedings, Vol. 51, 1955, pp.
1240) is used. To prevent ASR, 10 % silica fume by mass of 497516 and 785812.
cementitious material will be used [22,23]. [6] Chatterji, S., Mechanism of Alkali-Silica Reaction and Expan-
sion, Proceedings, 8th International Conference on Alkali-
Metakaolin Aggregate Reaction in Concrete, K. Okada, S. Nishibayashi, and
Metakaolin is a relatively recent material developed to prevent M. Kawamura, Eds., 8th ICAAR Local Organizing Committee, The
deleterious ASR in concrete [27]. It is derived from the weath- Society of Materials Science, Kyoto, Japan, 1989, pp. 101105.
ering of feldspar in granitic rock types and manufactured by [7] Moranville-Regourd, M., Products of Reaction and Petrographic
calcining the kaolin to a temperature range of 700800C. Examination, Proceedings, 8th International Conference on
After grinding to pozzolan fineness, it is used to replace up to Alkali-Aggregate Reaction, K. Okada, S. Nishibayashi, and M.
Kawamura, Eds., 8th ICAAR Local Organizing Committee, The
10 % of the portland cement with metakaolin [24].
Society of Materials Science, Kyoto, Japan, 1989, pp. 445456.
[8] Landgren, R., and Hadley, D. W., Surface Popouts Caused by
Lithium Salts Alkali-Aggregate Reaction, Portland Cement Association
McCoy and Caldwell [25] first reported the exceptional capa- Research and Development Bulletin RD 121, 13 pp.
bility of lithium salts to inhibit expansive ASR. They found that [9] Stark, D., The Moisture Condition of Field Concrete Exhibiting
small threshold quantities of any of several such salts could Alkali-Silica Reactivity, Durability of Concrete, Proceedings,
prevent expansion when used with high-alkali cement and Pyrex Second International Conference, Montreal, Canada, SP-126,
glass. Later work [26,27] verified these findings using highly Vol. II, American Concrete Institute, Detroit, MI, 1991, pp. 974
reactive commercially available aggregate. Here, it was 987.
recommended that lithium hydroxide (LiOH) salt be used [10] Mielenz, R. C., Greene, K. T., and Benton, E. J., Chemical Test
because it is soluble in tap water, unlike lithium fluoride or for Reactivity of Aggregates with Cement Alkalies: Chemical
lithium carbonate, thereby permitting better dispersal in large Processes in Cement-Aggregate Reaction, Journal, American
concrete batches. It also has been included in experimental Concrete Institute; Proceedings, Vol. 44, 1948, pp. 193219.
field concrete pavement. Currently, there are no specification [11] Stark, D., Alkali Silica Reactivity: Some Reconsiderations,
guidelines for its use, but the experimental laboratory work Cement, Concrete, and Aggregates, Vol. 2, No. 2, ASTM
International, West Conshohocken, PA, 1980, pp. 9294.
suggests a dosage rate of about 1:1 LiOH H2O: equivalent
[12] Rogers, C. A. and Hooton, R. C., Leaching of Alkalies in Alkali-
Na2O in the portland cement. Today, lithium nitrate is often Aggregate Reaction Testing, Proceedings, 8th International
used to reduce potential hazardous effects from the user [27]. Conference on Alkali-Aggregate Reaction in Concrete, K.
Okada, S. Nishibayashi, and M. Kawamura, Eds., 8th ICAAR
Local Organizing Committee, The Society of Materials Science, [21] Bakker, R. F. M., About the Cause of the Resistance of Blast-
Kyoto, Japan, 1989, pp. 327332. furnace Cement Concrete to the Alkali-Silica Reaction, Pro-
[13] Stark, D. and Bhatty, M. S. Y, Alkali Silica Reactivity: Effect of ceedings, Fifth International Conference on Alkali-Aggregate
Alkali in Aggregate on Expansion, STP 930, ASTM Interna- Reaction in Concrete, S252/29, NBRI of the Council for Scientific
tional, West Conshohocken, PA, 1985, pp. 1630. and Industrial Research, Pretoria, South Africa, 1981.
[14] Oberholster, R. E. and Davies, G., An Accelerated Method for [22] Gudmundsson, G. and Olafsson, H., Silica Fume in Concrete16
Testing the Potential Alkali Reactivity of Siliceous Aggregates, Years of Experience in Iceland, Proceedings of the 10th Inter-
Cement and Concrete Research, Vol. 16, 1986, pp. 186189. national Conference on Alkali-Aggregate Reaction in Concrete,
[15] Stark, D., Morgan, B. Okamoto, P., and Diamond, S., Eliminat- 1823 August 1996, Melbourne, Australia, pp. 562569.
ing or Minimizing Alkali-Silica Reactivity, Final Report, Strate- [23] Gundmundsson, G., Halfdanarson, J., and Mller, J., Effect of
gic Highway Research Program 343, 1993, 266 pp. Silica Fume Properties on Mitigation of ASR Reactivity in
[16] Stark, D., Immersion Test to Identify Cement Alkali Levels and Concrete, Alkali-Aggregate Reaction in Concrete, Proceed-
Pozzolans to Prevent ASR, Proceedings, 10th International ings, 11th International Conference, Quebec City, QC, Canada,
Conference on Alkali Aggregate Reactions in Concrete, June 2000, pp. 643651.
Melbourne, 1996, pp. 355361. [24] Sibbick, R. G., and Nixon, P. J., Investigation into the Effects of
[17] Shayan, Ahmad, Combined Effects of Alkali-Aggregate Reac- Metakaolin as a Cement Replacement in ASR Reactive Con-
tion (AAR) and Cathodic Protection on Currents in Reinforced crete, Alkali-Aggregate Reaction in Concrete, Proceedings,
Concrete, 11th ICAAR CIRAG, Quebec 2000, pp. 229238. 11th International Conference, Quebec City, QC, Canada, June
[18] Tarii, K., Kawamwa, M., Matsumoto, K., and Ishii, K., Influence 2000, pp. 763772.
of Cathodic Protection on Cracking and Expansion of the Beams [25] McCoy, W. J. and Caldwell, A. G., New Approach to Inhibiting
Due to Alkali-Silica Reaction, Proceedings, 10th International Alkali-Aggregate Reaction, Journal, American Concrete
Conference on Alkali Aggregate Reactions in Concrete, Institute, Proceedings, Vol. 47, 1951, pp. 693706.
Melbourne, 1996, pp. 355361. [26] Stark, D., Lithium Salt AdmixturesAn Alternative Method to
[19] Bhatty, M. S. Y., Mechanisms of Pozzolanic Reactions and Prevent Expansive Alkali-Silica Reactivity, Proceedings, 9th
Control of Alkali-Aggregate Expansion, Cement, Concrete and International Conference on Alkali-Aggregate Reactions in
Aggregates, Vol. 7, No. 2, ASTM International, West Concrete, London, 1992, pp. 10171025.
Conshohocken, PA, pp. 6977, 1985. [27] Folliard, K. J., Thomas, M. D. A., and Kurtis, K. E., Guidelines
[20] Stark, D., Considerations in the Use of Fly Ash to Control for the Use of Lithium to Mitigate or Prevent Alkali-Silica
Alkali-Silica Reactivity, Portland Cement Association, Serial Reaction (ASR), Publication No. FHWA-RD-03-047, July 2003,
No. 1987, 1995, 43 pp. 78 pp.
35
Alkali-Carbonate Rock Reaction
Michael A. Ozol1
Preface As demand for concrete increased, in large part to meet

the requirements of transportation-related construction, new
THE SUBJECT OF ALKALI-CARBONATE ROCK REAC- sources for aggregates were necessarily developed. It should be
tivity (ACR) was discussed for the first time as a distinct and sep- noted that new sources include previously unused or undevel-
arate chapter-length topic in ASTM STP 169B by H. N. Walker [1]. oped strata or portions of old operating limestone quarries as
Prior to that, W. C. Hansen, in ASTM STP 169A in 1966 [2], well as new greensite quarry sources. And, at around the
discussed the reactivity of dolomitic limestones, along with the same time, in the late 1960s and early 1970s, in response to
reactivity of other rocks and minerals, under the general heading of economic and environmental constraints, cement kiln dust,
chemical reactions. Previous editions of Significance of Tests and which is richer in alkalies than the cement product, began to
Properties of Concrete and Concrete-Making Materials be returned to the product. Thus, the preceding Factors 1 and 2
(ASTM STP 169, 1956) and its predecessors ASTM STP 22A, in no longer favored so much the satisfactorily innocuous behav-
1943, and ASTM STP 22, in 1935, did not treat the subject of dele- ior of aggregates with respect to alkalies in the cement, and in-
terious carbonate rock reactions because such reactions were not stances of deleterious behavior became more numerous.
recognized as a distinct problem until identified as such by Swen- son The following ASTM documents are directly applicable to
in 1957 [3], based on his investigations of expanded, cracked, and the subject of alkali-carbonate reactivity and are listed with a
buckled concrete in the vicinity of Kingston, Ontario, Canada. notation as to their particular relevance.
In addition to the now numerous occurrences of reactive
carbonate rock in concrete in Ontario, in the United States there ASTM Standard Specification for Concrete
have been affected concretes in Virginia, West Virginia, Ken- Aggregates (C 33)
tucky, Missouri, Tennessee, Iowa, Illinois, Indiana, and New York; Directs that deleteriously reactive aggregates shall not be used
and elsewhere in the world in Iraq, Bahrain, England, and China. in moisture-exposed concrete made with cement containing
more than 0.60 % total alkalies and, in the Appendix, describes
Introduction the petrographic characteristics of potentially reactive carbon-
ate rocks and, in part, how to evaluate them.
When a portland cement concrete mixture is designed and
placed, it is expected that the coarse and fine aggregate will not ASTM Guide for Petrographic Examination of
participate in any chemical reactions with components of the Aggregates for Concrete (C 295)
cement paste that will cause the aggregate particles to crack or States that petrographic examinations should identify and call
to become otherwise dimensionally unstable. attention to potentially alkali-carbonate reactive constituents,
Historically, there was a time when that expectation was determine their amounts, and recommend additional tests to
satisfactorily fulfilled, or at any rate believed to have been sat- determine whether those amounts are capable of deleteriously
isfactorily fulfilled, because for certain instances of concrete affecting concrete. Reference is made to examination of drilled
deterioration, the reasons for that deterioration were ascribed core and ledge rock. These topics are of particular importance
to causes other than deleteriously reactive aggregates. since it is more often the case that alkali-carbonate reactive rock
The satisfactorily innocuous, or inert, behavior of aggre- comprises a relatively small stratigraphic portion of a pro-
gates was, at that time, favored by such factors as: ducing quarry.
1. Continued availability of aggregate materials from old
sources, those materials that had performed without prob- ASTM Descriptive Nomenclature for
lems over years of consumption. Constituents of Natural Mineral
2. The general availability of low-alkali cement that was fur- Aggregates (C 294)
nished as a matter of course during the years when kiln dust Describes the characteristics, texture, and composition of car-
was not included in the product. bonate rocks that are potentially alkali-reactive, in essentially
the same language as is in ASTM C 33.
1
Consultant, geologist/photographer, Concrete and Materials, Baltimore, MD 21209.
410
OZOL ON ALKALI-CARBONATE ROCK REACTION 411
ASTM Test Method for Potential Alkali ASTM Method for Reducing Field Samples of
Reactivity of Carbonate Rocks for Concrete Aggregate to Testing Size (C 702)
Aggregates (Rock Cylinder Method) (C 586) Samples taken in accordance with ASTM D 75 may be reduced
Directly measures the expansion of small cylinders of rock dur- in size by quartering or by use of a mechanical splitter as de-
ing exposure in sodium-hydroxide solution. The rock sample is scribed in ASTM C 702.
secured in accordance with the applicable requirements of
ASTM D 75. Positive test results cannot, by themselves, be used Types of Carbonate Rock Reactions
as an accurate predictor of the damage that the rock will cause
in concrete. ASTM C 1105 (see next paragraph) should be used Although various reactions involving carbonate rock, both
to evaluate the behavior of the rock in concrete. dolomitic and nondolomitic, in concrete or mortar have been
recognized, only one, the expansive dedolomitization reaction,
which is the principal subject of this chapter, is of importance
ASTM Test Method for Length Change of with respect to having produced significant damage in con-
Concrete Due to Alkali-Carbonate Rock Reaction crete constructions. This reaction may or may not produce rim
(C 1105) zones on particles of reacting rock in concrete.
Directly measures the expansion produced in concrete prisms, Other reactions, causing the production of rim zones and
stored at 23C and 95 % RH, by the alkali-carbonate reaction, other internal changes in carbonate rocks of various composi-
using up to 19.0 mm (3/4 in.) maximum size aggregate, and the tions are of scientific interest. Some other reactions are dis-
job cement. cussed in Refs 1, 4, and 5. Katayama [6] gives a detailed review
of rim formation.
ASTM Practice for Examination and Sampling of
Hardened Concrete in Constructions (C 823)
Expansive Dedolomitization Reaction
Provides general and specific guidance for evaluating con-
The expansive dedolomitization reaction occurs between alka-
structions in the field and for obtaining samples for labora-
line solutions (almost always sodium and potassium hydrox-
tory examination as by ASTM C 856 (see next paragraph),
ides derived from the cement2) in the cement paste matrix and
with which it may be used in an interactive way, with ASTM
limestone aggregate particles of particular but nevertheless
C 856 providing preliminary information for detailed and rig-
somewhat variable compositions and textures. However, a fea-
orous application of ASTM C 823 to provide a second set of
ture common to all such limestones is that there will be small
samples for petrographic examination by ASTM C 856. ASTM
dolomite crystals present in the matrix of the rock that will re-
C 823 is of particular relevance when evaluating an aggregate
act with the alkaline solutions in a manner such that the dolomite
source for alkali-carbonate reactivity based on its service
crystals, of original composition CaMg (CO3)2, become other
record in previous constructions. In that connection, it is im-
minerals and compounds; hence, the term dedolomitization, as
portant to have information on the alkali content of the ce-
represented, simply, here, and as discussed in greater detail in
ment used, and the cement content of the concrete in the
subsequent sections.
constructions.
calcium magnesium carbonate (Dolomite)
ASTM Practice for Petrographic Examination of + alkali hydroxide solution magnesium hydroxide
Hardened Concrete (C 856) + calcium carbonate + alkali carbonate
Provides the general method for how to go about examining
The chief physical result of dedolomitization, which dis-
field and laboratory concrete chiefly microscopically to estab-
tresses and damages concrete, is that the reaction causes the af-
lish whether alkali-carbonate reactions have taken place, and
fected aggregate particles to expand and crack and thereby to
gives some specific criteria for recognizing the effects of alkali-
extend the cracks into the enclosing mortar matrix.
carbonate reactions.
Manifestations of Distress
ASTM Practice for Sampling Aggregates (D 75)
Gives the procedure for obtaining samples for examination Concrete that is affected by the alkali-carbonate rock reaction
and testing by ASTM C 295, C 1105, C 586, and any other expands and cracks. Depending on the perfection of the petro-
tests and examinations from which general conclusions re- logical characteristics of the rock that promote expansion, and
garding the aggregate may be drawn. It should be noted that the convergence of the environmental and materials-related
ASTM C 586 is best, and most conveniently, done on ledge factors that promote expansion, important among which are
rock samples that will permit the preparation of three mutu- availability of moisture and alkali content in the concrete, the
ally perpendicular small rock cylinders. Information regard- concrete may in a matter of months exhibit severe surface
ing obtaining ledge rock samples is in ASTM D 75. Alterna- cracking accompanied by closing of joints, as was reported in
tively, in the event that ledge rock samples cannot be Canada by Swenson [3]. Pattern cracking after three years of
obtained, it is possible that the small cores (1.38 in., 35 mm) exposure is shown in Fig. 1.
needed for ASTM C 586 can be drilled out of larger particles When all of the factors favoring alkali-carbonate reaction
for example, 2 in. (50 mm) or largerof aggregate obtained combine to produce near-ideal conditions for causing the reac-
from stock pile or conveyor belt. tion, spectacular expansion and deterioration, as reported by
2
Alkalies from deicing materials such as sodium chloride (NaCl) may augment the alkalies supplied by the cement.
Fig. 1Typical cracking of concrete slab by alkali-carbonate reaction after

three years, Ontario, Canada [7].
Rogers [8], can occur within three years. In that instance, the and the cement content of the concrete was about 300 kg/m 3
concrete in sidewalks, curbs, and gutters expanded about 1.2 % (500 lb/yd3).
(Fig. 2) with consequent compressive failures, blow-ups, and In ACR-affected field concrete, there is no definitive fea-
shoving of adjacent asphalt pavement. It is reported that the ture of the concrete or of the geometry of the cracking that
alkali content of the cement was about 1.0 % Na 2O equivalent, precisely identifies the cause of the cracking as being alkali car-
bonate-rock reaction. (In contrast, gel deposits, if present, may
provide a strong indication that cracking is due to alkali-silica
reaction but not exclude the possibility that both reactions can
be present.) The crack geometry, or pattern, produced is the re-
sponse of the concrete to internal expansive force based on its
particular size and shape and the direction of maximum
moisture availability. Generally, expansion of ordinary, plain,
cast-in-place concrete as in sidewalks, floors, decks, slabs, and
footings where there is a moisture gradient from top to bottom
or from side to side results in pattern cracking similar to that
shown in Fig. 1. Where moisture conditions are uniform or
where the concrete elements are thick, or of low surface area-
to-volume ratio, bulk expansion of the concrete is evidenced by
closing of joints, extrusion of filler material, shoving or buck-
ling of adjacent materials, or crushing of adjacent weaker con-
cretes. Cracks may be prominent or faintvisible only follow-
ing wetting of the surface to produce some contrastand
concrete between the cracks can be hard and intact.
Conclusive evidence that the observed distress has been
caused by alkali-carbonate reaction is best obtained by detailed
examination of the concrete following ASTM C 856, during
which it will be observed that the aggregate exhibits the
characteristic microscopic texture and composition of alkali-
reactive carbonate rock (see the section on Characteristics of
Alkali-Reactive Carbonate Rocks), and that internal cracking in
the aggregate extends into the mortar.
Reinforced and post-tensioned concrete, precast and pre-
stressed concrete, and other kinds of prefabricated concrete can
develop individualized cracking responses to internal ex-
pansion that may not be similar to those produced in slabs on
Fig. 2Concrete slab affected by alkali-carbonate grade. For example, elongate prestressed or post-tensioned
reaction (upper portion of photo) that has moved 9 cm to the pieces typically develop long cracks parallel to the direction of
right relative to the concrete in the lower portion of the photo tensioning or prestressing.
that has nonreactive aggregate. Expansion was measured at In at least one particular type of exposure condition, the
1.2 % in three years, Ontario, Canada [48]. manifestation of distress due to ACR does not involve gross ex-
pansion of the concrete and widespread cracking, but instead, the which, by precedent, have been regarded as prototypical. How-
damage is confined to the surface region and is the result of the ever, as will be noted subsequently, there are expansive,
application of deicing salts. For example, a parking deck investi- dedolomitizing, alkali-carbonate reactive rocks that do not con-
gated by the author containing reactive carbonate aggregate form to the historical prototype [13,14]. The classic, or proto-
throughout was deeply scaled exposing numerous cross-frac- typical, microscopic texture and structure is characterized by
tured coarse aggregate particles in the scaled surface. Only the relatively larger, rhombic crystals of dolomite [CaMg (CO 3)2]
coarse aggregate particles in the near-surface region had cracked, set in a finer-grained matrix of calcite [CaCO3], clay, and
extending cracks into the mortar to connect with similar cracks (commonly) silt-sized quartz. The characteristic composition is
from other reactive coarse aggregate particles, with the result that in which the carbonate-mineral components of the rock
that the surface concrete was lost. Although the deck has reactive consist of substantial amounts of both dolomite and calcite, and
aggregate throughout, the alkali content of the concrete, in the the dilute hydrochloric acid (HCl) insoluble residue contains a
absence of deicing salts, was insufficient to sustain the reaction significant amount of clay.
and to develop the overall expansion and tensile cracks produc- Examples of typical microstructures of dolomitic lime-
ing the map, or pattern cracking, that is usually observed. stones known to be alkali-carbonate reactive are shown in Fig. 3.
Smith [9] has presented experimental data showing the ex- The rhombic crystals of dolomite occurring in the matrix
acerbating effect of sodium chloride (NaCl) on alkali-carbonate of the reactive rock may be relatively sparsely distributed and
reactivity. Using concrete prisms, made with a 1.1 % alkali ce- appear to be floating in the background (Fig. 3, Field A) or
ment and a known reactive aggregate, stored in saturated NaCl may be more crowded together with dolomite rhombs adja-
solution at 70F (21C), he recorded about 0.19 % expansion at cent or touching (Fig. 3, Field E).
300 days. The same concrete stored in pure water at the same The grain sizes of the calcite and clay matrix in which the
temperature showed 0.12 % at 300 days. dolomite rhombs are set is typically 2 to 6 m for the calcite
Using reactive aggregate from the same quarry (Pittsburgh with smaller clay particles disseminated throughout. It is often
quarry at Kingston, Ontario) about 27 years later, Alasali et al. the case that silt-sized quartz grains are also disseminated
[10] recorded comparable results, measuring 0.23 % expansion throughout the matrix.
at 275 days on concrete prisms made with 1.13 % alkali cement The typical texture that is described here is found in those
stored in 5 % NaCl at 23C. A similar prism stored in water at reactive carbonate rocks that are identified as early expanders,
38C measured 0.155 % expansion at 275 days. showing rock prism expansions (ASTM C 586) of a few tenths
Gillott and Rogers [11] showed that alkali contributed by percent in a matter of weeks, and field manifestations in con-
certain rare types of mineral components in the coarse aggre- crete within perhaps one year after construction.
gate (e.g., dawsonite) can exacerbate the ACR. A modification of the typical texture is found in reactive car-
bonate rocks identified as late expanders that may not show no-
Comparison of Alkali Carbonate Rock Reaction ticeable rock prism test expansions until approximately 25
with Alkali-Silica Reaction as Regards weeks (rarely, not before one year) and that may not show man-
Generation of Crack Damage ifestations in concrete after five years in moist storage. In those
rocks, the typical feature of dolomite rhombs in a clayey fine-
In alkali silica reaction (ASR), the first damage to the concrete grained matrix is consistent, but the matrix is coarser-grained
can be produced by both (a) internal cracking of aggregate and is composed of interlocking dolomite grains together with
particles extending cracks into the paste and mortar, and (b) calcite, clay, and, more commonly than not, silica minerals.
extension of existing cracks and production of new cracks as a Those differences between early and late expanders are re-
consequence of the migration and subsequent expansion of flected in the bulk compositions, as shown in Table 1, and in dif-
the ASR gel, initially produced by the reacting particles, away ferences in internal textural restraint, or rigidity, as investigated
from the particles and into cracks and voids in the concrete. by Hilton [15,16] and discussed subsequently. Whether the pos-
In contrast, direct damage due to alkali carbonate reaction sibility of late (rock prism) expanders causing distress in field
is produced by the cracking, and extension of cracks into the mor- concrete after many years of exposure should be disregarded is
tar, generated within the reactive coarse aggregate particle itself. an open question.
The initial, or direct, damage produced in both ASR and ACR The above description of the alkali-carbonate reactive tex-
may then be exacerbated by the action of cyclical freezing and ture derives from the type locality in Ontario, Canada where
thawing. And, it should be recognized that, although bulk expan- the rocks were first identified and described. And, the descrip-
sions of concrete produced solely by the alkali-carbonate reaction tion has historically served well as a prototype for recognizing
may be lower than those produced in severe alkali-silica reaction, the ACR texture in North America and elsewhere. However, ex-
the ultimate damage to the concrete can be of the same order as tensive studies on ACR rocks in China [13] concluded that the
produced by ASR if the concrete is exposed in an environment prototypical texture was less frequently observed, due to the
where cyclical freezing and thawing occurs [12]. That is, the ini- variable character of the matrices, and that there are only two
tial cracks produced by the alkali-carbonate reaction are then ex- necessary petrographic conditions: the presence of fine-
ploited by saturation of the cracks and subsequent freezing. grained dolomite crystals and a pore structure allowing alkali
solution to penetrate the rock. Milanesi et al. [14] report on a
Characteristics of Alkali-Reactive very expansive, porous, fine-grained (0.010.03 mm) dolomite
Carbonate Rocks with 1020 % clay minerals but with no calcite.
The best technique for identifying reactive texture is by use
Petrographic of thin sections with a petrographic microscope.
The alkali-carbonate reactive limestones that were first studied Alternatively, the reactive texture can be identified in
in Canada, and shortly thereafter in the United States, have char- opaque section in concrete or rock samples, using a highly pol-
acteristic microscopic textures and mineralogical compositions ished specimen and manipulation of the illumination to get the
Fig. 3Photomicrographs of thin sections of alkali-carbonate reactive limestones from quarries

in eastern Canada. Scale bar is 0.25 mm. Fields A, B, and C are from different quarries. Fields D, E, and
F are from the same quarry, D and F from the same bed [48].
TABLE 1Composition of Early and Late Expansive

Carbonate Rocks (Adapted From Ref 1)
Dolomite
Acid Insoluble % of Total
Residue, % Carbonate
Kingston, Ontario, 5 to 15 about 50

early expanders [3,17,18]
Iowa, Illinois, and Indiana 10 to 20 40 to 60
early expanders [19,20]
Virginia early expanders [21,22] 13 to 29 46 to 73
Gull River, Ontario, 21 to 49 75 to 87
late expanders [18]
Virginia late expander [23] 33 >90
abundant in the overall category of siliceous aggregates

(crushed stone together with sand and gravel) that might be used
in concrete than are reactive carbonate rocks in the category of
all carbonate rocks. And, generally, but not entirely, the
carbonate aggregates that participate in alkali-carbonate reaction
are crushed stones from limestone quarries. This is in contrast to
alkali-silica reactivity where reactive siliceous components
can derive from both gravel and crushed stone sources. That is,
except near their bedrock source, and under conditions of rapid
erosion and deposition, limestone-bearing gravels, considering
earths gravels as a whole, are not as abundant as siliceous
gravels because limestones are less able than siliceous rocks to
survive the weathering and transportation necessary to put them
in gravel deposits as hard and sound particles suitable for use
as concrete aggregate. Consequently, alkali-carbonate reactive
particles in gravel deposits, because such particles are
limestones, are less likely to be found in gravels than alkali-
silica reactive particles. However, gravel deposits containing
Fig. 4Vein of reactive carbonate texture in limestone reactive carbonate rock exist and are reported to have been used
displaying alternating bands of reactive and nonreactive in concrete in England in the M 50 Motorway 3 and have been
texture. Northern Virginia, field is approximately 2 mm across. found and used in concrete, causing damage, in Ontario, Canada,
The same type of structure can also exist on a larger scale, from near their bedrock source.4
bed to bed, or within a bed, with the alternating bands being
several inches thick. Chemical and Mineralogical Composition
Typical compositions of early and late expanding alkali-reac-
proper contrast and reflectance. A light etching of the speci- tive expansive carbonate rocks are shown in Table 1, with their
men surface with dilute HCl can be very helpful. geographic location and publication reference indicated.
Carbonate rocks displaying reactive texture can comprise In an extended study of the composition of reactive and
a substantial or a small portion of a bed in a limestone quarry. nonreactive Ontario limestones and their expansion in the Rock
Or, the reactive texture may occur as veins within a limestone Cylinder and Concrete Prism Expansion Tests, Rogers [8]
bed that does not otherwise display the reactive texture (Fig. 4). found that the compositions of potentially expansive rocks fell
The occurrence of the reactive carbonate rock texture is a within a fairly distinct field when the CaO:MgO ratio is plotted
reliable diagnostic guide. Rocks which exhibit the texture against either the Al2O3 content or the insoluble residue. The
should be tested further to determine their expansive potential plot versus insoluble residue is shown in Fig. 5. Those empirical
with the contemplated job cement. relationships are the basis for the Canadian Standard Chemical
Alkali-carbonate reactive coarse aggregate particles in Method screening test for ACR, CSA A23.2-26A.
concrete may or may not display reaction rims. The formation
of reaction rims on ACR rock particles is incidental to their Factors Affecting Expansion of Concrete with
propensity for reaction and generation of crack damage. Reactive Carbonate Rock and Factors
However, the presence of reaction rims calls for closer inspec- Affecting Expansion of Reactive Carbonate
tion of those coarse aggregate particles, since often, but not al- Rock by Itself
ways, their presence is indicative of alkali-carbonate reaction.
Broadly speaking, it appears to be the case geologically that A suite of carbonate rocks that have reactive texture and that ex-
the conditions for formation of alkali-expansive carbonate rocks hibit varying degrees of (unrestrained) expansion in the rock
are more restricted than for other limestones, and they are not cylinder test (ASTM C 586) may exhibit a very different expansion
abundant in the volume of limestone that exists on earth. They ranking when tested in restraint in the rock cylinder test [15,16]
may occur in carbonate rocks of any age but, thus far, have been or when restrained by incorporation in concrete [23]. Based on
found most often in carbonate rocks of Ordovician age. Hiltons work [15] the degree to which reactive carbonate rocks
It is clear from the microtexture of reactive carbonate will cause expansion in concrete is related, on the one hand, to
rocks, with the commonly sharp-edged rhombs, and occur- the restraint imposed by the concrete, and, on the other, to the
rence as veins of reactive texture, that that texture is of re- volume of dolomite in the rock, up to a point, and the internal
placement, or possibly diagenetic, origin rather than one textural restraint, or rigidity, of the carbonate rock in question.
resulting from primary deposition. As such, it is reasonable to A reactive carbonate rock with low textural restraint can be
infer that, since the production of the reactive texture requires highly expansive unrestrained, but when restrained, either
special conditions subsequent to deposition and not obtaining experimentally or within concrete, it is more compressible and
as components of the ordinary depositional regime for lime- less expansive.
stones, there will be less of it in the total limestone sectionas In Hiltons experimental work [15,16], the most expansive
field experience thus far seems to indicate. rock, unrestrained in ASTM C 586, expanded 7.8 % at 14 weeks
To put reactive carbonate rocks in perspective with reactive and, restrained in a steel frame in 1N sodium hydroxide
siliceous rocks, reactive siliceous rocks are far more
3
W. J. French, personal communication.
4
C. A. Rogers, personal communication.
Fig. 5CaO:MgO ratio versus insoluble residue for a suite of alkali-carbonate reactive
and nonreactive carbonate rocks [8].
(NaOH), expanded 0.275 % at 14 weeks. The rock has 58.6 % by <6 % calcite, and 0.18 % porosity. It did begin to expand after
volume dolomite, 23.5 % calcite, and 5.63 % porosity. Petro- 40 weeks of continuous soaking in the NaOH solution.
graphically, the rock has a high volume of dolomite rhombs Qian et al. [24] obtained similar results using autoclaved
tightly packed but with abundant voids between the masses of concrete microbars (150C in 10 % KOH solution) incorporat-
tightly packed rhombs (Fig. 6). ing mosaic-textured, almost pure, crystalline dolomites. They
The least expansive sample did not expand at all after eight concluded that these rocks do undergo dedolomitization but that
weeks in 1N NaOH. It is a typical crystalline dolomite com- early expansion is very low compared to rocks with the typical
posed of a mosaic of equant dolomite grains in a tight virtually ACR texture. However, with sufficiently long reaction time
nonporous structure (Fig. 7). It had 94 % dolomite by volume, significant expansion could develop.
Fig. 6Expansive rock from Missouri, 7.8 % in 14 weeks in ASTM C 586. High
content of dolomite rhombs in porous matrix.
in cement pores; (5) high proportion of reactive stone in the

coarse aggregate; and (6) lower concrete strength [23,25].
Mechanism of Reaction and Expansion

Following the first recognition of the alkali-carbonate reaction
by Swenson [3], work by Hadley [4], and later investigators,
showed that the main chemical reaction that occurred in the
rock was that the dolomite [CaMg(CO3)2] decomposed, or
dedolomitized, to calcite (CaCO 3) and brucite Mg(OH)2, as
represented by the following reaction in which M represents an
alkali element, such as potassium, sodium, or lithium:
CaMg(CO3)2 + 2MOH Mg(OH)2 + CaCO3 + M2CO3
Fig. 7Nonexpansive rock, in ASTM C 586, from Virginia. A further reaction that occurs is that the alkali carbonate
Structural framework of equant dolomite grains with no dis- produced in the initial reaction may then react with the
crete floating rhombs and no interstitial matrix. Ca(OH)2, produced as a normal product of cement hydration,
to regenerate the alkali hydroxide. For example:
Further in Hiltons work, a rock, representative of several, Na2CO3 + Ca(OH)2 2 NaOH + CaCO3
with the typical alkali-carbonate reactive texture (Fig. 8) and
similar to Fig. 3 (Fields A, B, and F), expanded unrestrained Although the starting and end products of the reaction
1.4 % at 14 weeks, and restrained 0.31 %. It has 32.4 % by vol- have been well documented and establish the overall bulk
ume dolomite, 36.7 % calcite, and 1.25 % porosity. chemical and mineralogical changes that take place, the
The rock with the lesser value of unrestrained expansion dedolomitization reaction, as such, does not translate into a
expanded more under restraint than the rock with five times mechanical explanation for the expansion, because the volume
greater unrestrained expansion, and almost two times greater of the reaction products is smaller than that of the reactants
dolomite volume. [26]. The explanation was sought in behavior of the reactive
Thus, carbonate rocks with different or equivalent chemi- rock and alkali system with respect to indirect mechanisms
cal/mineralogical composition could have vastly different and secondary reaction products that had been identified in
expansion characteristics in concrete if their structural fabrics various investigations. That is, the primary dedolomitization
differ to such an extent that their elastic properties are quite reaction was considered to be the prerequisite that enabled
different. other processes to work to produce the expansion.
The petrological factors leading to the optimization of the An idea common to some proposed mechanisms is that, in
lithology for producing expansion in concrete are: (1) clay con- one way or another, water absorption and incorporation is im-
tent, or insoluble residue content, in the range of 525 %; (2) portant in producing the expansion.
calcite to dolomite ratio of approximately 1:1; (3) increase in Sherwood and Newlon [27] proposed that expansion could
dolomite volume up to the point at which interlocking texture be related to the formation of the hydrated, relatively high-
becomes a restraining factor; and (4) small size of the discrete volume, low-density, secondary mineral products gaylus- site
floating dolomite rhombs. and buetschliite. Feldman and Sereda [28] found trace amounts,
The expansion of concrete containing alkali-reactive car- in reacted rock, of a material that expands when it ab- sorbs
bonate rocks is increased by: (1) increasing coarse aggregate water. Walker [29] and Buck [30] found minerals of the
size; (2) moisture availability; (3) higher temperature; (4) high hydrotalcite-sjogrenite group, which may possibly help pro-
alkali content of the concrete and high pH of the liquid phase mote water absorption and consequent expansion.
Swenson and Gillott [31] proposed that dedolomitization
exposes active clay minerals that are present as inclusions, in
the rocks that they studied, within the dolomite euhedra. Ex-
change sites on the clay surface adsorb sodium ions, and water
uptake by the new clay results in swelling. Thus, it is the ac-
tive clay released during dedolomitization that contributes to
the expansion.
Clay inclusions within the dolomite rhombs of expansive
rocks have not been universally reported. They have not been
observed, for example, in expansive rocks from Virginia.
Tang et al. [32] conclude that the expansion due to ACR is
caused by ions and water molecules migrating into restricted
spaces created by the growth and rearrangement of the products
of dedolomitization, particularly brucite. Therefore, the
expansion results directly from dedolomitization and not from
water uptake of clay mineral inclusions within dolomite
Fig. 8Expansive rock from northwest Virginia with
typical reactive texture. Expansion in ASTM C 586 was 1.4 % at rhombs. Using STEM and EDAX to observe the locations of the
14 weeks.
reaction products in expanded samples of Kingston, Ontario, the formation of brucite, the uptake of water, and the formation
reacted rock, they found that the brucite produced by the re- of iron-rich phases from ferroan dolomite.
action occurred in a 2-m-thick ring of parallel-oriented 25 Tong [13] states that it is feasible for brucite to react with
crystals, with space between them, surrounding the euhedral silica in the alkaline environment, to form gel-like materials,
dolomite rhombs of the typical reactive texture. The brucite and that it is therefore sensible that it would not be detected.
crystals were, in turn, surrounded by the calcite and clay par- He notes also that brucite can be difficult to detect due to the
ticles of the matrix. The calcite of the dedolomitized rhomb re- small amount produced and its poor crystallinity.
mained within the interior of the dedolomitized crystal. Prince and Perami [38] showed that dedolomitization is
Later work, discussed below, using compacts of cement and possible in the absence of alkali by experiments on pastes com-
dolomite, or magnesite, to simulate ACR rocks, is considered posed of dolomite mineral powder, soda, calcium chloride, and
by Tong, Tang, and others to confirm this expansion mechanism lime, in various proportions. The pastes were cured 24 h at 40C
[26,33,34]. in a saturated atmosphere and the reaction products monitored
The role of the clay minerals in the fabric of the rock may by XRD. In paste composed of only dolomite and lime, only
be to provide mechanical pathways to the dolomite rhombs by brucite and calcite were formedthe dedolomitization
disrupting the structural framework of the rock, thus weakening occurred in the absence of any alkali. With introduction of soda,
the carbonate matrix and, by that means, helping to promote dedolomitization was enhanced. The general conclusion was
expansion. Thus, the presence of clays is not essential to the that the particular reaction products depend on the ratio of
chemistry of the dedolomitization reaction but is an enabling calcium to sodium ions. High values of the ratio promote the
factor, mechanically promoting expansion when formation of brucite and calcite, while lower values promote
dedolomitization occurs. brucite and pirssonite [CaNa2 (CO3)2H2O].
Katayama [6], based largely on a review of the literature, Milanesi et al. [14] related expansion of concrete prisms
suggests that there is a stronger correlation between expansion and rock cylinders to dedolomitization in both types of samples
and silica (insoluble residue) content of carbonate rocks having and in powdered samples of the same three rocks immersed in
ACR texture than with their dolomite content, and that strong alkaline solutions, using XRD, SEM, and optical mi-
deleterious expansion is confined to impure carbonate rocks that croscopy to identify the reaction products. The most expansive
contain more than 510 % insoluble residue. Therefore, alkali- rock in both the rock cylinder test (ASTM C 586 but using 1975
carbonate reaction should be reviewed in light of the alkali- mm cylinders) and the concrete prism test, CSA A23.2-14A, was
silica reactivity of cryptocrystalline quartz hidden in the a porous, fine-grained, dolomite with a minor amount of quartz,
microstructure of the matrix enclosing the dolomite rhombs. feldspar, and rhyolite clasts, but with no evident calcite. Previ-
For several Chinese aggregates, Tong and Tang [35] found ous work showed the rock to be definitely not alkali-silica reac-
that both alkali-reactive dolomite and alkali-reactive silica, as tive. Examination of the cylinder before and after immersion in
microcrystalline quartz, are present in the same aggregate par- 1 N NaOH solution showed strong dedolomitization had oc-
ticle. Both dedolomitization and silica gel formation occurred in curred, producing calcite and brucite. A calcitic dolomite and a
rapidly expansive concrete microbars cured at either 60C or dolomitic limestone, both of coarser grain size and lower poros-
150C in 10 % KOH solution. XRD and SEM/EDAX results ity than the expansive rock, shrank in the rock cylinder test and
confirmed the presence of the products of both reactions. expanded only about 6 to 7 % of the 5-year prism expansion
Investigators have noted that brucite may not always be (0.205 %) of the most expansive sample. Those samples showed
detectable as evidence that dedolomitization has occurred in no evidence of dedolomitization in the rock cylinder test. Pow-
concrete that has expanded [13,3537]. The reasons for this are ders of the three rocks completely dedolomitized in 1 N NaOH
discussed subsequently. after four days, forming calcite and brucite with no detectable
Experiments aimed at establishing and reconfirming the dolomite. The authors attribute the differences in expansion of
chemistry and products of the dedolomitization reaction, and the the three dolomitic rocks to differences in their microstructural
expansion associated with the reaction, have been conducted characteristics. They conclude that the equations proposed by
using single crystals of mineral dolomite, dolomite mineral Hadley [4] are adequate to explain dedolomitization in concrete
powders, cylinders of dolomitic rocks, and mortar compacts of under normal service conditions.
powdered, high-purity dolostone, and of magnesite, with Tang et al. [34] and Tong and Tang [26,33] using mortar
portland cement. compacts of portland cement with high-purity, powdered dolo-
Randonjic et al. [37] confirmed the basic dedolomitization stone and high-purity, powdered magnesite immersed in alka-
reaction by examining single crystals of dolomite, before and af- line solutions at elevated temperatures performed experiments
ter immersion in NaOH solution, using environmental scanning aimed at addressing the following questions: (1) whether the
electron microscopy (ESEM), atomic force microscopy (AFM), volume-reducing alkali-carbonate reaction (the volume of the
and the surface analytical technique X-ray photoelectron spec- products being less than the volume of the reactants) can cause
troscopy (XPS). After immersion in NaOH they observed the expansion; (2) whether absorption of water by clay exposed by
chief reaction products of calcite and brucite and that the first dedolomitization causes the expansion; (3) whether the ACR is
reaction product to form was brucite, followed by calcite. Their actually an alkali-silica reaction due to the reactivity of cryp-
results suggest that the brucite would be susceptible to further tocrystalline quartz hidden in the matrix enclosing the dolomite
chemical change by reacting with species from the alkaline en- rhombs; (4) whether the presence of the matrix of calcite, clay,
vironment such as Ca2+ and Si4+. And, that this should also be and fine-grained quartz enclosing the dolomite rhombs in the
true for magnesium and iron rich phases formed during prototypical ACR texture is a necessary condition for
dedolomitization. The authors propose that since the chemical expansion by dedolomitization.
changes occurring during dedolomitization involve surface re- The use of compacted cylinders of dolomite or magnesite
actions, they would be expected to involve expansive volume particles in a matrix of cement paste simulates the spatial rela-
changes. Processes which may contribute to expansion include tion of dolomite rhombs enclosed in a fine-grained matrix, as
is the case in natural alkali-reactive dolomitic limestones, and crete was low alkali cement, less than 0.6 % total alkalies as per
eliminates the effects of cryptocrystalline quartz in the matrix, ASTM Specification for Portland Cement (C 150), then the con-
clay in, or between, the dolomite rhombs, and calcite, clay, and crete cannot provide information on the propensity for reaction
quartz comprising the matrix, focusing instead on the reaction of the aggregate with high alkali cement. However, in view of
of the dolomite or magnesite particles in the cement paste. the facts, (1) the most reactive carbonate rocks can produce
Magnesite was used as a cross reference material [26,33] deleterious expansion with cement of 0.4 % total alkalies [23],
because it also reacts to produce brucite in an expansive reac- and (2) the more important measure is the alkali content per unit
tion in which the brucite is the unique solid product, simplify- volume of concrete, not the alkali content of the cement per se.
ing the expansion mechanism, and it is also a solid volume Then even if the cement used was less than 0.6 % total alkalies,
reducing process. if the cement content of the concrete was high, the alkali level
Compacts were also made with limestone powder [34], high per unit volume may equal or exceed that of concrete in which
purity alumina, and cement paste by itself, as control samples. the cement content was low but the alkali content of the cement
In addition to the expansions of the compacts, the expansions was high. A threshold guideline, based on pounds of sodium
of rock prisms and concrete microbars of the same materials oxide (Na2O) equivalent per cubic yard, must be established
were measured. Reaction products were determined by X-ray for judging whether the aggregate should be considered to have
diffraction. been used in a high or low alkali concrete.
The control samples produced negligible expansions while If the structure shows no distress and it has been estab-
the magnesite and dolomite compacts exhibited considerable lished that the alkali content of the concrete would be suffi-
expansions which correlated well with the dolomite consump- cient to cause reaction if the aggregate is potentially reactive,
tion/calcite production and the magnesite consumption/brucite then the effects of aggregate size, proportion of reactive ag-
production in the respective compacts. The authors conclude gregate, and moisture availability to the concrete must also be
that the results provide a powerful confirmation that the ex- considered singly and in combination before concluding that the
pansion was directly caused by the reactions occurring at the aggregate is non-deleteriously expansive.
interfaces of the dolomite or magnesite particles with the en- Larger-sized reactive aggregate causes greater expansion
closing cement paste matrix. And that, although it is a solid than smaller-sized reactive aggregate. A reactive aggregate that
volume reducing reaction, an expansive force is generated by has not caused distress with high alkali cement in small top
the growth and rearrangement of the reaction products, to- size may cause distress when used in the larger size with the
gether with the migration of alkali and hydroxyl ions, and water, same cement.
into the confined space of the particle/matrix interface. The greater the proportion of reactive aggregate of the to-
It is reasonable that the compacts are a good, functional, tal aggregate, the greater the expansion. A reactive aggregate
proxy, artificial aggregate for the ACR limestone in which that has not caused distress as a small portion of the coarse ag-
dolomite crystals are enclosed in a fine-grained matrix. The con- gregate in a structure with high alkali level may cause dis-
clusion may then be drawn from the experiments that neither cal- tress when used in larger proportion with the same cement.
cite, silica, or clay in the matrix, nor clay in, or between, the Potentially reactive aggregates will not cause deleterious
dolomite rhombs is related to the expansion-causing process. amounts of expansion if the concrete is protected and in a rela-
Rather, the expansion is directly proportional to the dedolomiti- tively dry condition. Where exposure to moisture also includes
zation of the dolomite enclosed in the matrix and must, therefore, exposure to sodium-chloride-containing deicing salts, which are
be related to the accumulation and positioning of the reaction known to exacerbate ACR, the conclusion that the aggregate is
products in relation to the surface of the dolomite rhombs. non-deleteriously expansive is strengthened.
Evaluating the Potential for Alkali Concrete Prism Expansion Test

Carbonate Reactivity The surest measure of the potential for deleterious expansion of
a carbonate aggregate in concrete is the measurement of the
Field Service Record length change (for percent expansion) of concrete prisms, as by
Field performance information on concrete incorporating the ASTM C 1105 or Canadian Standard CSA A23.2-14A.
aggregate under consideration can furnish direct answers to ASTM C 1105 is intended primarily for evaluating a par-
the questions that the laboratory evaluation tests attempt to an- ticular cement-aggregate combination prior to use in concrete,
swer indirectly. That is, tests on the rock, in or out of concrete, although, as noted, it can be used for investigational or research
are intended to provide information on which, in essence, to purposes (Paragraphs 4.5 and 6.3).
base a prediction on what will happen if the rock is used in con- When it is used for its primary purpose the alkali content of
crete construction. However, if good information is available the cement will be that of the job cement, and the storage con-
from the field, then, logically, that is best, and indirect infor- ditions will be at 23C and > 95 % relative humidity. If the al-
mation to predict what can be observed directly is second best. kali content is raised by using a higher alkali cement and/or
Provided certain essential information, noted later, is avail- adding NaOH to the mixing water, and the storage temperature
able and certain exposure conditions are fulfilled, an evaluation is raised, the test is accelerated and is then generally similar to
of the potential for aggregate reactivity based on the investiga- CSA A 23.214A and ASTM C 1293, in which the storage
tion of concrete structures in the field can be the most unequiv- temperature is 38C (100.4F) and the cement content is stipu-
ocal and economic way of accomplishing the evaluation. lated at 420 kg/m3 (708 lb/yd3), with a total alkali content, using
When investigating and sampling a concrete structure cement of 0.9 % alkali augmented by NaOH in the mixing wa-
(ASTM C 823) known to contain aggregate from the source in ter, of 1.25 % by mass of cement, or, 5.25/kg/m3 (8.85 lb/yd3).
question, for example, by prior petrographic examination, the C 1105 is closer to field conditions than is CSA A 23.2-14A
alkali content of the cement and the cement content per cubic which is intended to accelerate the expansion by maximizing the
yard must be known. In general, if the cement used in the con- conditions for the detection of slowly reactive rocks and for
TABLE 2Concrete Prism Test, Expansion Limits, and Alkali Content

Expansion Limit
at Indicated Time, Percent
Na2O Equivalent
Test Alkali Content 3 months 6 months 1 year
ASTM C 1105 job cement 0.015 0.025 0.030

CSA A23.2-14A 1.25% ... 0.040
facilitating comparisons among aggregates by using exactly the The expansion limit of 0.040 % used in the CSA standard
same aggregate grading and alkali level. is the approximate level at which cracking and distress are first
In C 1105 the grading of the job-proposed aggregate can observed on the prisms, and it relates well to observed field
be used if it is not coarser than 19 mm (3/4 in). But, C 1105 performance [39].
specifies that if the job-proposed aggregate has material larger
than 19 mm (3/4 in.), it shall not be used, based on the as- Rock Cylinder Expansion Test
sumption that the plus 19-mm material does not differ in com- The expansion of small (approximately 1.4 in. in length by 0.4
position, lithology, and presumed expansive characteristics in. in diameter, 35.5 by 10 mm) cylinders of carbonate rock that
from the minus 19-mm material. If, based on information from are continuously immersed in 1 N NaOH solution has been used
petrographic examination or rock cylinder tests (ASTM C 586) as a direct indication of expansive alkali-carbonate reactivity of
or both, the plus 19-mm material differs from the minus 19- the rock since the method, originally using square cross section
mm material in those respects, then instructions are given for specimens, was first described and used by Hadley in 1961
testing the coarser fraction. In CSA A23.2-14A, the coarse ag- [20]. And since its earliest use, the correlation between rock
gregate is separated on 20-, 14-, 10-, and 5-mm sieves, oversize cylinder (or rock prism) expansion and concrete prism
and undersize material discarded, and equal masses of the expansion has been substantiated (Figs. 9 and 10). How- ever,
three sizes recombined. note in Fig. 10 the spread of the six rock prism expansion values
Expansion limits for ASTM C 1105 have not been adopted corresponding to the concrete prism expansion value (average
in ASTM C 33 but are suggested in the Appendix to ASTM C 1105 of three) and, in Fig. 9, that the rock prism expansion values
and are, therefore, nonmandatory. As such, they place greater plotted are the average of 20 specimens.
responsibility on the specifier of the concrete to understand the Thus, despite good correlations between the average val-
factors promoting expansion due to alkali-carbonate reactivity ues of rock prism and concrete prism expansions, due to the
versus the service environment of the concrete. variability inherent in the rock, the differences that can exist
The expansion limit for the CSA prism test is specified and between restrained (in concrete) and unrestrained (in ASTM C
tabulated in Table 2 with the suggested ASTM C 1105 limits. 586) behavior, and the other factors governing expansion in
It may be noted with respect to, for example, the three-month concrete (for example, w/c, total water soluble alkalies, paste to
C 1105 limit, that fifteen-thousandths of one percent is not a dele- aggregate ratio, etc.), the situation is such that ASTM C 586 is
terious degree of expansion. But the logic of the test is such that,
based on correlation with longer term tests and field perform-
ance, a rock which produces 0.015 % expansion at three months
will produce deleterious expansion in field concrete at a later age.
Fig. 10Relationship between one-year length change of

(ASTM C 157) concrete prisms (average of three) with high
Fig. 9Expansion of small rock prisms in alkaline solution alkali cement (0.95 %) and length change of (ASTM C 586)
correlates well with expansion of concretes using these rocks rock prisms at eight weeks. Average and range of six rock
as aggregate [4]. prism length changes are shown for each sample [23].
a relatively rapid indicator of potential expansive reactivity, Petrographic examination speaks directly to the concerns
and an effective tool for screening aggregate sources, but it of Paragraphs 10.1, 10.2, and 10.3, Interpretation of Results
does not predict in-concrete expansion of the rock and is not of ASTM C 586. And, although not specifically called for in
used as an acceptance test. ASTM C 1105, petrographic examination of the concrete
One test specimen per sample of rock is the minimum re- prisms at the conclusion of the test can furnish information
quired by the method. It should be taken normal to the bed- helpful for interpretation of results at, or near, the test limits.
ding if bedding is discernible. If bedding is not discernible, The Canadian Standard notes the usefulness of petrographic
then three mutually perpendicular specimens are made. In examination at this stage.
general, even if bedding is discernible, it is desirable to prepare
three mutually perpendicular specimens; one perpendicular to Determination of Potential Alkali
the bedding and two at right angles parallel to the bedding. Of Carbonate Reactivity of Carbonate Rocks
three such specimens, the one exhibiting the greatest expan- by Chemical Composition
sion is the one on which to base the test result. The Ontario, Canada, Ministry of Transportation and Commu-
An expansion of 0.10 % at 28 days is considered to indicate nications has developed a quick chemical screening test for the
the presence of material with a potential for deleterious ex- determination of potential alkali-carbonate reactivity of car-
pansion when used in concrete. For example, as reported by bonate rocks by chemical compositionCanadian standard CSA
Rogers [8], most quarry faces sampled in Ontario, Canada, that A 23.2-26A.
gave average rock cylinder expansions of greater than 0.1 % at The test is based on work by Rogers [8] and uses relation-
four weeks also caused excessive expansion and cracking when ships based on a correlation between expansion of concrete
used in concrete. prisms and rock cylinders, and performance of aggregates in
Although 0.1 % expansion at 28 days is the implied limit field structures, and the CaO:MgO ratio of the limestone versus
given in ASTM C 586, an alternate limit of 0.20 % at 16 weeks is the alumina (Al2O3) content or the insoluble residue content.
also considered applicable by several researchers in view of the For example, in Fig. 5, using insoluble residue, regions are
fact that some rocks contract before they begin to expand delineated in which the potentially expansive carbonate rocks
[23,40]. For example, a delayed expander tested by Newlon, et al. cluster. Rocks with compositions in these regions should be
[23] contracted 0.1 % at four weeks but expanded to 0.25 % at 16 considered to be potentially deleterious until their nondelete-
weeks, eventually peaking at 0.5 % at 44 weeks. A typical alkali rious level of expansion is demonstrated by service record or by
carbonate reactive rock producing on the order of 0.1 % concrete a concrete prism expansion test.
prism expansion at one year with high alkali cement might
expand on the order of 0.5 % at four weeks in ASTM C 586. Other Methods and Approaches
In the suite of sample test results reported by Rogers [8]
there were considerably fewer samples in the category < 0.1 % Concrete Microbars
at four weeks but > 0.2 % at 16 weeks than there were in the Xu et al. [41] developed an accelerated method for determining
category > 0.1 % at four weeks. potential alkali-carbonate reactivity using concrete microbars
In view of the requirements of specimen selection and the 202080 mm and 4040160 mm stored in 1 N NaOH
occasional irregular, or late expanding, behavior of samples in solution at 80C for four weeks. The aggregate particle size
the rock cylinder test, it is strongly recommended that all as- used is 510 mm, with cement (boosted by addition of KOH) of
pects of the testing program and interpretation of the results 1.5 % Na2O eq. at a w/c of 0.30 and aggregate to cement ratio of
be guided by petrographic information. 1:1. An expansion of 0.1 % at four weeks is judged to be ap-
propriate for recognizing ACR limestones, based on comparing
Petrographic Examination the behavior of aggregates in the test and in field concrete.
Petrographic examination of dolomitic carbonate rocks Grattan-Bellew et al. [42] propose a modification of the
(ASTM C 295) proposed for use as coarse aggregate in con- microbar test, using 4040160 or 286 mm bars, for alkali-
crete will detect the presence of the characteristic reactive carbonate reactive aggregates [41] which is also applicable to
lithologic texture. And, the characteristic reactive texture, as alkali-silica reactive aggregates. They state also that siliceous
described in ASTM C 294 and discussed elsewhere in this ASR limestones can be differentiated from dolomitic ACR lime-
chapter, when it occurs in a dolomitic limestone, will undergo stones by running two tests, one with 100 % portland cement,
reaction or dedolomitization in an alkaline environment. the other replacing a portion of the cement with 30 % fly ash. If
Whether, when tested, the reaction will produce expansion of the limestone is alkali-silica reactive the expansion will be sig-
a rock cylinder (ASTM C 586) or concrete beam (ASTM C nificantly reduced, whereas the fly ash will have little effect on
1105), insofar as the rock is concerned, depends on the the expansion of the ACR limestone aggregate.
perfection of the type lithology, as regards ASTM C 586, and as
regards ASTM C 1105, on the amount and distribution of the Selection of Samples for Testing and for
reactive lithology within the stratigraphic section proposed for Concrete Aggregate
use as concrete aggregate.
Following the initial use of petrographic examination for Quarry Sampling
detecting the reactive carbonate rock texture (see Quarry Sampling of the stratigraphic section in limestone quarries for
Sampling) in order to identify candidate samples for further detecting the presence of alkali-carbonate reactive rock by pet-
testing by ASTM C 586 or C 1105, petrographic examination is rographic examination should recognize that the characteristic
valuable during successive steps of the evaluation to guide the reactive texture and composition are only subtly and indirectly
understanding of the behavior of samples in, and the interpre- related to the primary factors of sedimentation that produce
tation of results from, the rock cylinder test (ASTM C 586) and bedding. The reactive texture is of secondary, or replacement,
the concrete prism test (ASTM C 1105). origin and may or may not conform faithfully to lithologic or
stratigraphic units, or individual beds as observable in the field. mind that cement with total Na2O equivalent alkali at the 0.4 %
The amount of reactive texture and the expansivity of that tex- level can produce alkali-carbonate reaction, and that alkali con-
ture may vary laterally and vertically within beds or within a se- tent per unit volume of the concrete is the important measure.
quence of beds comprising a definable, or mappable, strati- Therefore, the proposed job concrete mixture should be tested
graphic unit, to the same extent that the variations in rock prism by ASTM C 1105.
(ASTM C 586) expansions within a given bed, or lithology, may Unlike the alkali-silica reaction, the alkali carbonate re-
be as large or larger than those between lithologies [23]. action is not very well controlled by the use of fly ash or ground
Sampling for the reactive lithology should therefore be granulated blast furnace slag (GGBFS). Tang et al. [34] found
guided by petrographic examination using as many thin or pol- that even with cement of 0.43 alkalies, expansion of concrete at
ished sections as necessary to delineate the reactive rock and one year, cured at 40C and 100 % RH, was effectively reduced
to select samples for further tests. And, because of the varia- only at 70 % replacement of cement by fly ash, or 90 % by slag.
bility that can exist, any thin or polished section that shows Rogers and Hooton [43] in a five year comparison study of
even a minor area of reactive texture should, at the first sam- laboratory and field concrete with and without GGBFS
pling stage (Stage I), require the field sampling unit that it rep- concluded that up to 50 % replacement of high alkali cement
resents to be considered suspect [23]. (1.04 %) with GGBFS did not reduce expansion sufficiently so
The objective of the first (Stage I) sampling is to establish that the Kingston, ON, reactive aggregate could be safely used.
the presence or absence of reactive rock within a quarry. It is Thomas and Innis [44], using the same aggregate in concrete
understood that the establishment of 100 % absolute absence prism tests (CSA A23.2-14A) and up to 65 % replacement of
requires examination of every bed. Lithology and structure high (1.25 %) alkali and 50 % replacement of low (0.50 %) alkali
should be considered in a way that samples are located with due cement, found that the slag did not effectively control
regard to their relationship to the quarry production horizons, expansion.
and in sufficient detail to permit a calculated weighting for Also, unlike the ASR, the use of lithium compounds does
comparison with the maximum allowable reactive content of not mitigate the expansion caused by ACR. Results of Wang and
the total aggregate that can be established by subsequent test- Gillott [45], using the Kingston, ON, reactive aggregate, showed
ing using ASTM C 1105. In Stage I sampling, a minimum of five that concrete prisms (CSA A23.2-14A) with either LiCL or
and an average of approximately ten samples should be taken in LiOH, at nine months, had the same expansion as the control
a quarry. Each lithology representing more than 10 % of the prism, and the prism with LiNO3 had significantly more ex-
production should be sampled. Lithologies representing less pansion than the control. Stark [46] recommends that the pre-
than 10 % can be combined and sampled at random locations in ferred compound for ASR mitigation is LiOH due to its greater
the combined thickness. Where the face exhibits relatively thin solubility than either lithium fluoride or lithium carbonate. Tang
beds of varying lithology, it should be sampled at a minimum et al. [47], using the Kingston, ON, aggregate in concrete bars
of ten systematically located stations [25]. treated at 150C in water and in 10 % LiOH and KOH, obtained
If Stage I sampling establishes the presence of reactive rock the greatest expansion for the KOH, intermediate for the LiOH,
within a quarry, then the objective of the Stage II sampling will and lowest for the water (control) sample. The authors note that
be to identify, as precisely as the limitations of time and budget LiOH produced dedolomitization and brucite formation, similar
will permit, the precise beds and their amounts of reactive ma- to the reactions with Na and K hydroxides, and that the use of
terial. Perhaps from 50 to 100 samples may be required. Previ- LiOH could be an effective means to dif- ferentiate alkali-
ous experience suggests that the results of the more extensive carbonate from alkali-silica reaction.
and finer-scale sampling can end up delineating a lesser volume In the use of the concrete in constructions that will con-
of reactive material in the quarry face than may have been in- tain diluted, or otherwise minimized, amounts of reactive
dicated from the results of the Stage I sampling [25]. aggregate and alkali, restrict availability of moisture to the con-
crete by barriers or by design of the construction if possible.
If no mitigating approach is feasible, prohibit the use of the
Using Coarse Aggregate from a Quarry with
reactive rock in portland-cement concrete.
Potentially Expansive Rock
If potentially expansive rock is present in a quarry, the follow- Acknowledgment
ing courses of action should be considered.
If the stratigraphy of the producing section is favorable, The author is grateful to Christopher A. Rogers, Ministry of Trans-
avoid the reactive rock by selective quarrying. If the reactive portation, Downsview, Ontario, Canada, who kindly provided im-
rock cannot be totally avoided, utilize selective quarrying such portant graphs, photographs, and technical information.
that the aggregate from the produced section, when tested in
concrete by ASTM C 1105, does not produce a deletrious ex- References
pansion. If the natural dilution approach is infeasible, dilute
the produced aggregate with nonreactive stone such that the [1] Walker, H. N., Chemical Reactions of Carbonate Aggregates in
resulting mixture does not produce a potentially deleterious Cement Paste, Significance of Tests and Properties of Concrete
and Concrete-Making Materials, ASTM STP 169B, ASTM Inter-
level of concrete expansion by ASTM C 1105.
national, West Conshohocken, PA, 1978, pp. 722743.
Research and experience with a group of Virginia quarries
[2] Hansen, W. C., Chemical Reactions, Significance of Tests and
indicated that, for those sources, an acceptable level of dilution
Properties of Concrete and Concrete-Making Materials, ASTM
was reached when the reactive rock did not exceed 20 % of the STP 169A, ASTM International, West Conshohocken, PA, 1966,
coarse aggregate, 20 % of the fine aggregate, or 15 % of the to- pp. 487496.
tal aggregate, if both fractions contained reactive material [25]. [3] Swenson, E. G., A Reactive Aggregate Undetected by ASTM
In the concrete, use the smallest aggregate size that is eco- Tests, Bulletin No. 226, ASTM International, West Con-
nomically feasible and the lowest alkali cement, bearing in shohocken, PA, 1957, pp. 4851.
[4] Hadley, D. W., Alkali Reactivity of Dolomitic Carbonate Rocks, [22] Sherwood, W. C. and Newlon, H. H., Jr., A Survey for Reactive
Proceedings, Symposium on Alkali-Carbonate Rock Reactions, Carbonate Aggregates in Virginia, Record No. 45, Highway
Highway Research Board, Record No. 45, 1964, pp. 120. Research Board, Washington, DC, 1964, pp. 222233.
[5] Mather, B., Developments in Specification and Control, [23] Newlon, H. H., Ozol, M. A., and Sherwood, W. C., Potentially
Cement-Aggregate Reactions, Transportation Research Record Reactive Carbonate Rocks, An Evaluation of Several Methods
525, Transportation Research Board, Washington, DC, 1974, for Detecting Alkali-Carbonate Reaction, Progress Report No. 5,
pp. 3842. Virginia Highway Research Council, 71-R33, Charlottesville, VA,
[6] Katayama, T., A Critical Review of Carbonate Rock Reactions May 1972.
Is Their Reactivity Useful or Harmful, Conference Papers of the [24] Qian, G., Deng, M., and Tang, M., ACR Expansion of Dolomites
9th International Conference on Alkali-Aggregate Reaction in with Mosaic Textures, Magazine of Concrete Research, Vol. 53,
Concrete, 1992, Vol. 1, pp. 508518, Concrete Society Publica- No. 5, 2001, pp. 327336.
tion CS. 104, U.K. [25] Newlon, H. H., Jr., Sherwood, W. C., and Ozol, M. A., Poten-
[7] Field Trip Guidebook, 7th International Conference on Alkali- tially Reactive Carbonate Rocks, Progress Report No. 8., A
Aggregate Reaction, Ontario Ministry of Transportation, Engi- Strategy for Use and Control of Carbonate Rocks Including an
neering Materials Office, Materials Information Report 99, Annotated Bibliography of Virginia Research, VHRC 71-R41,
Toronto, Canada, 1986. Virginia Highway Research Council, Charlottesville, VA, June
[8] Rogers, C. A., Evaluation of the Potential for Expansion and 1972.
Cracking of Concrete Caused by the Alkali-Carbonate [26] Tong, L. and Tang, M., Expandability of Solid-Volume-Reducing
Reaction, Cement, Concrete, and Aggregates, Vol. 8, No. 1, Reactions of Alkali-Magnesite and Alkali-Dolmite, Cement,
Summer 1986, pp. 1323. Concrete, and Aggregates, Vol. 19, No. 1, 1997, pp. 3137.
[9] Smith, P., Learning to Live with a Reactive Carbonate Rock, [27] Sherwood, W. C. and Newlon, H. H., Jr., Studies on the Mech-
Proceedings, Symposium on Alkali Carbonate Rock Reactions, anisms of Alkali-Carbonate Reaction, Part I. Chemical Reac-
Record No. 45, Highway Research Board, 1964, pp. 126133. tions, Highway Research Record No. 45, Highway Research
[10] Alasali, M. M., Malhotra, V. M, and Soles, J. A., Performance of Board, Washington, DC, 1964, pp. 4156.
Various Test Methods for Assessing the Potential Alkali [28] Feldman, R. F. and Sereda, P. J., Characteristics of Sorption
Reactivity of Some Canadian Aggregates, ACI Materials and Expansion Isotherms of Reactive Limestone Aggregate,
Journal, Vol. 88, No. 6, Nov.-Dec. 1991, pp. 613619. Journal, American Concrete Institute; Proceedings, Vol. 58, No.
[11] Gillott, J. E. and Rogers, C. A., Alkali-Aggregate Reaction and 2, 1961, pp. 203214.
Internal Release of Alkalis, Magazine of Concrete Research, [29] Walker, H. N., Reaction Products in Expansion Test Specimens
Vol. 46, No. 167, 1994, pp. 99112. of Carbonate Aggregate, Transportation Research Record 525,
[12] Ozol, M. A. and Newlon, H. H., Jr., Bridge Deck Deterioration National Research Council, Washington, DC, 1974.
Promoted by Alkali-Carbonate Reaction: A Documented Exam- [30] Buck, A. D., Control of Reactive Carbonate Rocks in Concrete,
ple, Cement-Aggregate Reactions, Transportation Research Technical Report C-75-3, U.S. Army Engineer Waterways
Record 525, Transportation Research Board, National Research Experiment Station, Corps of Engineers, Vicksburg, MS, 1975.
Council, Washington, DC, 1974, pp. 5563. [31] Swenson, E. G. and Gillott, J. E., Alkali-Carbonate Rock
[13] Tong Liang, Alkali-Carbonate Rock Reaction, Ph.D. Thesis, Reaction, Proceedings, Symposium on Alkali-Carbonate
Nanjing Institute of Chemical Technology, China, January, 1994. Rock Reactions, Highway Research Record No. 45, Highway
[14] Milanesi, C. A., Marfil, S. A., Batic, O. R., and Maiza, P. J., Research Board, Washington, DC, 1964.
The Alkali-Carbonate Reaction and its Reaction Products an [32] Tang, M., Liu, Z., and Han, S., Mechanism of Alkali-Carbonate
Experience with Argentinean Dolomite Rocks, Cement and Reaction, Concrete Alkali-Aggregate Reactions, Proceedings,
Concrete Research, Vol. 26, No. 10, 1996, pp. 15791591. 7th International Conference, 1986, Ottawa Canada, P. E.
[15] Hilton, M. H., Expansion of Reactive Carbonate Rocks under Grattan-Bellew, Ed., Noyes Publications, Park Ridge, NJ, 1987.
Restraint, Cement-Aggregate Reactions, Transportation [33] Tong, L. and Tang, M., Correlation Between Reaction and
Research Record 525, Transportation Research Board, National Expansion of Alkali-Carbonate Reaction, Cement and Con-
Research Council, Washington, DC, 1974, pp. 921. crete Research, Vol. 25, No. 3, 1995, pp. 470476.
[16] Hilton, M. H., The Effects of Textural and External Restraints [34] Tang Mingshu, Deng Min, Lan Xianghi, and Han Sufen,
on the Expansion of Reactive Carbonate Aggregates, Progress Studies on Alkali-Carbonate Reaction, ACI Materials Journal,
Report No. 7b, Virginia Highway Research Council, Char- Vol. 91, No. 1, January-February 1994, pp. 2629.
lottesville, VA, Sept. 1968. [35] Tong, L. and Tang, M., Concurrence of Alkali-Silica and Aklali-
[17] Swenson, E. G. and Gillott, J. E., Characteristics of Kingston Car- Dolomite Reaction, Alkali-Aggregate Reaction in Concrete,
bonate Rock Reaction, Concrete Quality Control, Aggregate Proceedings of the 10th International Conference on Alkali-
Characteristics and the Cement Aggregate Reaction, Bulletin No. Aggregate Reaction in Concrete, A. Shayan, Ed., August 1996,
275, Highway Research Board, Washington, DC, 1960, pp. 1831. Melbourne, Australia, pp. 742749.
[18] Dolar-Mantuani, L. M. M., Expansion of Gull River Carbonate [36] Tong, L. and Tang, M., A Case Study of Two Airport Runways
Rocks in Sodium Hydroxide, Proceedings, Symposium on Al- Affected by Alkali-Carbonate Reaction, Part Two: Microstruc-
kali-Carbonate Rock Reactions, Record No. 45, Highway Re- tural Investigations, Cement and Concrete Research, Vol. 27,
search Board, Washington, DC, 1964, pp. 178195. No. 3, 1997, pp. 329396.
[19] Hadley, D. W., Alkali-Reactive Carbonate Rocks in Indiana [37] Radonjic, M., Allen, G., Ragnarsdottir, V., and Livesey, P.,
A Pilot Regional Investigation, Proceedings, Symposium on Novel Techniques in a Study of the Mechanism of Dedolomiti-
Alkali-Carbonate Rock Reactions, Record No. 45, Highway zation in Alkali-Carbonate Reaction, Alkali-Aggregate Reac-
Research Board, Washington, DC, 1964, pp. 196221. tion in Concrete, Proceedings, 11th International Conference,
[20] Hadley, D. W., Alkali Reactivity of Carbonate Rocks 2000, Quebec City, Canada, pp. 99108.
Expansion and Dedolomitization, Proceedings, Highway [38] Prince, W. and Perami, R., Mechanism of the Alkali-Dolomite
Research Board, Vol. 40, 1961, pp. 462474. Reaction, Conference Papers of the 9th International Confer-
[21] Newlon, H. H. and Sherwood, W. C., An Occurrence of Alkali- ence on Alkali-Aggregate Reaction in Concrete, 1992, Vol. 2,
Reactive Carbonate Rock in Virginia, Carbonate Aggregate pp. 799806, Concrete Society Publication CS. 104.
and Steam Curing of Concrete, Bulletin No. 355, Highway [39] Thomas, M. D. A., Hooton, R. D., and Rogers, C. A., Prevention
Research Board, Washington, DC, 1962, pp. 2744. of Damage Due to Alkali-Aggregate Reaction (AAR) in
Concrete ConstructionCanadian Approach, Cement, [44] Thomas, M. D. A. and Innis, F. A., Effect of Slag on Expansion
Concrete and Aggregates, Vol. 19, No. 1, 1997, pp. 2630. Due to Alkali-Aggregate Reaction in Concrete, ACI Materials
[40] Ryell, J., Chojnacki, B., Woda, G., and Koniuszy, Z. D., The Journal, Vol. 95, No. 6, November/December 1998, pp.
Uhthoff Quarry Alkali Carbonate Rock Reaction: A Laboratory 716724.
and Field Performance Study, Transportation Research Record [45] Wang, H. and Gillott, J. E., Alkali-Carbonate Reaction: Signifi-
525, Transportation Research Board, National Resarch Council, cance of Chemical and Mineral Admixtures, Magazine of
Washington, DC, 1974, pp. 4354. Concrete Research, Vol. 47, No. 170, 1995, pp. 6975.
[41] Xu, Z., Lan, X., Deng, M., and Tang, M., A New Accelerated [46] Stark, D., Lithium Salt AdmixturesAn Alternative Method to
Method for Determining the Potential Alkali-Carbonate Prevent Expansive Alkali-Silica Reactivity, Proceedings, 9th
Reactivity, Cement and Concrete Research, Vol. 32, 2002, International Conference on Alkali-Aggregate Reactions in
pp. 851857. Concrete, London, 1992, pp. 10171025.
[42] Grattan-Bellew, P. E., Cybanski, G., Fournier, B., and Mitchell, L., [47] Tang, M., Deng, M., and Xu, Z., Comparison Between Alkali-
Proposed Universal Accelerated Test for Alkali-Aggregate Silica Reaction and Alkali-Carbonate Reaction, Proceedings,
Reaction: The Concrete Microbar Test, Cement, Concrete, and 11th International Conference on Alkali-Aggregate Reaction in
Aggregates, Vol. 25, No. 2, 2003, pp. 2934. Concrete, Quebec, 2000, pp. 109118.
[43] Rogers, C. A. and Hooton, R. D., Comparison Between Labora- [48] Rogers, C. A., Alkali Aggregate Reactions, Concrete Aggre-
tory and Field Expansion of Alkali-Carbonate Reactive gate Testing and Problem Aggregates in OntarioA Review,
Concrete, Conference Papers of the 9th International Confer- Ontario Ministry of Transportation and Communications,
ence on Alkali-Aggregate Reaction in Concrete, 1992, Vol. 2, Engineering Materials Office, Report EM-31, 5th rev. ed.,
pp. 877884, Concrete Society Publication CS. 104. 1985.
36
Thermal Properties of Aggregates
D. Stephen Lane1
Preface The attention that has been given to the thermal proper-
ties of concrete aggregates is not as great as that given to other
THE CHAPTER ON THERMAL PROPERTIES OF properties of aggregates. This is probably because in many in-
concrete aggregates was authored by H. K. Cook in the first three stances the effects do not appear to be as important as the ef-
editions of Significance of Tests and Properties of Con- crete fects of other properties of aggregates and, consequently, the
and Concrete Aggregates, and by the current author in the fourth. significance of thermal properties is not as apparent. The need
Because little has changed in this subject area since the publication for a better understanding of the effects of the thermal
of ASTM STP 169C, this chapter is substantially the same as in characteristics of aggregates, particularly the coefficient of
the previous edition, with only minor revisions and additions to thermal expansion on the durability of concrete, has been
the text and reference list. expressed by Allen [1], Woolf [2], Scholer [3], and others.
The discussion in this paper is primarily confined to the
Introduction significance of tests and thermal properties of the aggregates.
While this necessarily requires discussion of the thermal prop-
Aggregates typically compose 6570 % or more of the volume erties of the concrete as affected by the aggregates, the signifi-
of concrete. Because the thermal properties of composite ma- cance of tests and thermal properties of concrete is discussed in
terials such as concrete are dependent on the thermal properties Chapter 24 of this publication.
of their constituents, the aggregates have a great influence on the
thermal properties of concrete. The degree to which thermal General Concepts of Thermal Properties
properties of concrete aggregates are of concern to the user
depends upon the nature of the concrete structure and the Thermal properties of rocks are related directly to the thermal
exposure to which it is subjected. Knowledge of thermal properties of the constituent minerals in proportion to their
properties are important in the design of massive structures, relative abundance. A reasonably good approximation of the
such as dams, where thermal and volume stability of the value of thermal expansion, specific heat, conductivity, or
composite are important; in concretes exposed to extreme tem- diffusivity of a rock can be calculated if the mode (mineral
peratures or temperature cycles, where the relative properties of content, percent by volume) of the rock is known using the
the individual constituents may be important; and in light- following formula [4]
weight concrete structures where insulating value of the com-
posite is a primary factor. Pt(rock) = S(n*1 Pt1 + n*2 Pt2 + n*3Pt3...) (1)
The thermal properties of aggregates that are of general
concern are the coefficient of thermal expansion, conductiv- where
ity, diffusivity, and specific heat. Of these, the thermal expan-
sion of aggregates is by far the most important because of the n1 = percent by volume, Mineral 1;
importance of the thermal expansion of concrete in the de- Pt1 = thermal property value of Mineral 1; and
sign of all concrete structures. The conductivity, diffusivity, S = solidity (1 decimal porosity).
and specific heat of aggregates are usually of concern only in
special instances. The most obvious of these is massive struc- The mode of an aggregate can be determined petrograph-
tures where the dissipation of heat generated by cement hy- ically using ASTM Guide for Petrographic Examination of
dration is important to control thermal cracking. Other in- Aggregates for Concrete (C 295). The value thus calculated for
stances include insulating concrete and the durability of a given aggregate will not be as accurate as could be measured.
concrete exposed to cyclical temperature changes. The design However, given the considerable care and skill necessary to
engineer should carefully evaluate the circumstances of a accurately measure these properties, a calculated value may
particular concrete application to estimate the importance that prove acceptable for many general applications. Robert- son [4]
thermal properties of aggregates will have on concrete has compiled a large volume of data on the thermal properties of
performance. rocks and minerals.
1
Research Scientist, Virginia Transportation Research Council, Charlottesville, VA 22923
OZOL
425 ON ALKALI-CARBONATE ROCK REACTION 426
TABLE 1Average Linear Coefficient of

Thermal Expansion of Some Common Rocks
and Minerals [5]
Mineral 106/C
Quartz 11.5 to 12
Orthoclase, microcline 6.5 to 7.5
Pyroxenes, amphiboles 6.5 to 7.5
Olivine 6 to 9
Albite 5 to 6
Calcite 4.5 to 5
Oligoclase, andesine 3 to 4
Labradorite, Bytownite 3 to 4
Anorthite 2.5 to 3
Rock 106/C
Quartzite, silica shale, chert 11.0 to 12.5

Sandstones 10.5 to 12.0
Quartz sands and pebbles 10.0 to 12.5
Argillaceous shales 9.5 to 11.0
Dolomite, magnesite 7.0 to 10.0
Granites and gneisses 6.5 to 8.5
Syenite, andesite, diorite, phonolite, 5.5 to 8.0
gabbro, diabase, basalt
Marbles 4.0 to 7.0
Dense, crystalline, porous limestones 3.5 to 6.0
Coefficient of Thermal Expansion

Numerical Values for Thermal Coefficients of Expansion
The coefficient of thermal expansion of aggregates is directly dependent on the coefficients of the
constituent minerals and thus on the mineral composition of the rock. Zoldners [5] has reported
coefficients of thermal expansion for common rock- forming minerals and some common rocks
that are presented in Table 1. Although values for the thermal coefficients of expansion of
aggregates from specific locations are contained in many of the references listed and are available
from many other sources, Refs 416 list more values than most of the others.
As will be noted from the table, quartz has the highest coefficient of thermal expansion of the
common minerals, while calcite and high calcium feldspars exhibit the lowest. Thus, the coefficients
of rocks vary to a large degree in direct proportion to their quartz content.
While most minerals will exhibit some degree of anisotropy by expanding more in a direction
parallel to one crystallographic axis than another, this is generally not of significance except when
the differences are great. The most no- table example is calcite, which can have a linear thermal coef-
ficient of expansion as great as 25.8 106/C parallel to its C-axis and as low as 4.7 106/C
perpendicular to this direction. Potassium feldspars are another group of minerals exhibiting rather
extreme anisotropy
For this reason, the cubical expansion of rocks and minerals is not always related directly to
the linear expansion, and this possibility should be kept in mind when investigating the thermal
properties of aggregates.
The paper by Meyers [17] is concerned completely with thermal coefficient of expansion of
portland cement rather
than aggregates. It is included because a discussion of the thermal properties of concrete
aggregates would not be complete without some information on the thermal properties of the
cements with which they are used. The coefficient of thermal expansion of hydrated cement
pastes may range from 10.8 106/C to 16.2 106/C and mortars from about 7.9
106/C to 12.6 106/C. Mitchell [7] has found that at early ages or at certain critical
saturations the linear thermal coefficient of expansion of cement pastes may be somewhat
higher than those reported earlier. He reports values as high as
22.3 106/C for some samples of neat cement specimens. An average value for the linear
thermal coefficient of ex-
pansion of concrete may be taken as 9.9 106/C, but the range may be from about 5.8
106/C to 14.0 106/C, depending upon the type and quantities of the aggregates, the
mixture proportions, and other factors.
Effect of Thermal Expansion

The significance of the coefficient of thermal expansion of aggregates with respect to its effect
on concrete is two-fold. The primary, or at least better understood, effect is that on the volume
change of concrete. Aggregates with high coefficients of thermal expansion will produce
concretes of low thermal volume stability and vice versa, all other things equal. The thermal
expansion of concrete should be accounted for in the design of the structure and is further
discussed in Chapter 24 on Thermal Properties in this publication. The other effect is the develop-
ment of internal stresses as a result of large differences between the coefficients of thermal
expansion of the various components of concrete subjected to temperature extremes or thermal
cycling and the impact this may have on concrete durability.
There seems to be fairly general agreement that the thermal expansion of the aggregate
has an effect on the durability of concrete, particularly under severe exposure conditions or
under rapid temperature changes. There is less agreement on whether more durable concrete
will be produced with an aggregate of high thermal coefficient of expansion similar to that
of the cement paste or with an aggregate of low thermal coefficient.
In one of the earliest papers on the subject, Pearson [18] attributes a case of rapid
concrete deterioration to the use of an aggregate of low thermal coefficient. The aggregate,
a dolomitic marble with a coefficient of thermal expansion of 3.6 106/C was reported to be
very resistant to freezing and thawing deterioration. The concrete was severely cracked after
the first winter, having been subjected to very low temperatures and to severe frost action.
Given the aggregates freezing and thawing durability and because the deterioration occurred
much more rapidly than would have been expected from freezing and thawing action alone,
other causes were suspected. Pearson speculated that tensile stresses developed when the matrix
contracted three to four times that of the coarse aggregate at extremely low winter temperatures.
These stresses were sufficient to cause fine cracking of the concrete providing pathways for
water into the concrete and subsequent freezing and thawing damage. Pearson conducted a
laboratory investigation [19] with concretes containing aggregates of both low and high thermal
coefficients. The concretes were subjected to 100 thermal cycles between 28.9 and 21.3C
before being subjected to freezing-and-thawing tests. Concretes containing the aggregates of
low thermal coefficient failed much more rapidly in the freezing-and-thawing test than the
concretes containing aggregates of high thermal coefficient.

Alkali-Silica Reactions in Concrete: David Stark

Uploaded by

Copyright:

Available Formats

Alkali-Silica Reactions in Concrete: David Stark

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Alkali-Silica Reactions in Concrete: David Stark

Uploaded by

Copyright:

Available Formats

34

Alkali-Silica Reactions in Concrete

Fig. 1Fine pattern or random cracking initiated by ASR in a parapet wall of a

Fig. 2Longitudinal cracking initiated by ASR in a jointed, slab-on-grade

Fig. 4Evidence of ASR in sample of concrete from gravity dam. Coarse

known to exacerbate alkali-silica reactivity, whereas restraint Use Admixed Materials

The mechanism by which slags inhibit expansive reactivity

Preface As demand for concrete increased, in large part to meet

Fig. 1Typical cracking of concrete slab by alkali-carbonate reaction after

Fig. 3Photomicrographs of thin sections of alkali-carbonate reactive limestones from quarries

TABLE 1Composition of Early and Late Expansive

Kingston, Ontario, 5 to 15 about 50

abundant in the overall category of siliceous aggregates

in cement pores; (5) high proportion of reactive stone in the

Mechanism of Reaction and Expansion

CaMg(CO3)2 + 2MOH Mg(OH)2 + CaCO3 + M2CO3

Evaluating the Potential for Alkali Concrete Prism Expansion Test

TABLE 2Concrete Prism Test, Expansion Limits, and Alkali Content

ASTM C 1105 job cement 0.015 0.025 0.030

Fig. 10Relationship between one-year length change of

Thermal Properties of Aggregates

TABLE 1Average Linear Coefficient of

Quartzite, silica shale, chert 11.0 to 12.5

Coefficient of Thermal Expansion

Effect of Thermal Expansion

You might also like